key: cord-0697586-239rkbfo
authors: Verna, Elizabeth C; Landis, Charles; Brown, Jr, Robert S; Mospan, Andrea R; Crawford, Julie M; Hildebrand, Janet S; Morris, Heather L; Munoz, Breda; Fried, Michael W; Reddy, K Rajender
title: Factors Associated with Readmission in the US Following Hospitalization with COVID-19
date: 2021-05-20
journal: Clin Infect Dis
DOI: 10.1093/cid/ciab464
sha: ac1a4eeff2dbd8dcd2d6f4d071b4648f628cdab4
doc_id: 697586
cord_uid: 239rkbfo

BACKGROUND: Patients hospitalized for COVID-19 may experience complications following hospitalization and require readmission. This analysis estimates the rate and risk factors associated with COVID-19-related readmission and inpatient mortality. METHODS: This is a retrospective cohort study utilizing deidentified chargemaster data from 297 hospitals across 40 US states on patients hospitalized with COVID-19 February 15-June 09, 2020. Demographics, comorbidities, acute conditions, and clinical characteristics of first hospitalization are summarized. Mulitvariable logistic regression was used to measure risk factor associations with 30-day readmission and in-hospital mortality. RESULTS: Among 29,659 patients, 1,070 (3.6%) were readmitted. Readmitted patients were more likely to have diabetes, hypertension, cardiovascular disease (CVD), chronic kidney disease (CKD) vs those not readmitted (p<0.0001) and to present on first admission with acute kidney injury (15.6% vs. 9.2%), congestive heart failure (6.4% vs. 2.4%), and cardiomyopathy (2.1% vs. 0.8%) (p<0.0001). Higher odds of readmission were observed in patients age >60 vs. 1840 (odds ratio [OR]=1.92, 95% confidence interval [CI]=1.48, 2.50), and admitted in the Northeast vs. West (OR=1.43, 95% CI=1.14, 1.79) or South (OR=1.28, 95% CI=1.11, 1.49). Comorbidities including diabetes (OR=1.34, 95% CI=1.12, 1.60), CVD (OR=1.46, 95% CI=1.23, 1.72), CKD stage 1-5 (OR=1.51, 95% CI=1.25,1.81) and stage 5 (OR=2.27, 95% CI=1.81, 2.86) were associated with higher odds of readmission. 12.3% of readmitted patients died during second hospitalization. CONCLUSIONS: Among this large US population of patients hospitalized with COVID-19, readmission was associated with certain comorbidities and acute conditions during first hospitalization. These findings may inform strategies to mitigate risks of readmission due to COVID-19 complications.

The COVID-19 pandemic has led to large numbers of hospitalizations and deaths throughout the world. A total of 124,006 laboratory-confirmed COVID-19-associated hospitalizations were reported to the CDC through the COVID-19-Associated Hospitalization Surveillance Network March 1, 2020-January 16, 2021,[1] while the overall cumulative COVID-19-associated hospitalization rate was 380.3 hospitalizations per 100,000 population. Inpatient mortality rates throughout the world have varied significantly, though data from the United States have revealed that over 20% of patients did not survive hospital admission in some large cohorts. Several demographic and clinical risk factors for both hospitalization and inpatient mortality have been identified, including advanced age, race, and comorbid conditions such as hypertension, diabetes, and chronic end organ disease. [2] [3] [4] It is likely that survival of the initial hospitalization fails to capture the true morbidity of severe forms of COVID-19. A significant proportion of patients are hospitalized for prolonged periods with numerous end organ complications of the virus itself or complications of critical illness.

Many patients who survive are discharged to hospice or skilled nursing facilities with variable degrees of debilitation. In addition, the long-term consequences of the COVID-19 syndrome remain largely undescribed, but may be significant. [5] Hospital readmission has emerged as an important quality indicator in health care delivery. [6, 7] Patients admitted for COVID-19 may have an increased risk of readmission given that they often experience one or more of the risk factors associated with readmission as documented in national datasets including: advanced age, medical comorbidities, socioeconomic factors, prolonged length of stay and skilled nursing facility residence. [8] Quantifying the rate of readmission for COVID-19 and identifying risk factors associated with readmission is crucial in providing accurate prognosis as well as appropriate therapies and treatments for preventing readmission-associated morbidity and mortality. [9] [10] [11] The aim of this study was to estimate the rate of readmission among patients A c c e p t e d M a n u s c r i p t discharged with confirmed COVID-19, and identify risk factors for readmission to target for intervention.

Data for these analyses were acquired from the hospital chargemaster, a comprehensive electronic claims database of all billable products, procedures, and services incurred during hospitalization. [4] As described previously, deidentified hospital chargemaster data consist of coding recorded by experts in accordance with the criteria established by Centers for Medicare and Medicaid

Services. [12] based upon review of each patient's hospital chart at or near discharge. These data were iteratively analyzed by a multidisciplinary team including a coding specialist, physicians, data scientists, an epidemiologist, scientist, and statistician. [4] [13] By incorporating the detailed information included in billing, procedure, and ICD-10 codes, as well as text searches, specific variables were created to allow for analysis of the conditions and journeys of patients hospitalized for Each institution used a common framework with consistent processing to compile data. A unique token was generated for each patient in order to ensure protection of Protected Health Information (PHI); this was compliant with the independent expert determination of compliance allowed for under 45 CFR Part 164.514(b). These tokens can be used to trace the data back to the original data sources and are part of the dataset used for this research. All parties assert that (a) the traceability of the data as a chain of custody from Target RWE to the individual institutions is maintained, and (b) that commercialization and licensing of said data for this use is permitted. [4] [13]

Patient demographics for analysis were: age (18-40, 41-60, >60), sex, insurance type, census region of hospital and type of hospital.

Comorbidities present on each admission and from hospital encounters within 12 months up to the date of admission were ascertained from ICD-10 codes and classified by expert opinion consistent with standard definitions commonly used in health services research. [14] All chronic and acute conditions were binary variables with a 'yes' value indicating that the condition was documented in the patient's claim records.

Factors related to disease severity and clinical care were assessed for the first hospital admission using billing codes, ICD-10 codes, and text searches as previously described. [4] Factors such as length of stay, ICU admission, level of oxygen support, presence of pneumonia and discharge disposition were assessed.

The primary and secondary endpoints were 30-day readmission after discharge from initial hospitalization for COVID-19 associated illness and death during the second hospitalization,

A c c e p t e d M a n u s c r i p t

Patient and hospitalization characteristics corresponding to first hospitalization were examined according to readmission status. Bivariate associations were assessed using chi-square tests for categorical data (Fisher's Exact for sparse data) and Kruskal-Wallis test for continuous data.

Multivariable logistic regression was used to estimate the adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the outcomes of interest in relation to age, sex, census region of hospital, insurance, comorbidities, length of stay, and complications developed during hospitalization.

To ensure that results were not biased toward selection of patients with shorter time to relapse or complications requiring readmission, all analyses were repeated excluding patients with <30 days follow-up after their initial discharge. All statistical tests were 2-sided with type 1 error (alpha)=0.05. SAS version 9.4 was used for all analyses. SAS JMP® 15.1.0 was used to produce forest plots.

A total of 30,054 patients were admitted between February, 15 and June 9, 2020 with COVID-19; 380 patients <18 years old were excluded from this analysis as were 15 adults with unknown sex (Figure 1) . The final analytic cohort comprised 29,659 adult inpatients with confirmed COVID-19, of whom 1,070 (3.6%) were readmitted within 30 days of first hospitalization discharge (Table 1) . Patients who were readmitted were significantly more likely to be older than age 60 years compared to those not readmitted (69.7% vs. 53.7%, respectively, p<0.0001), more likely to have Medicare or Medicaid-based insurance (67.5% vs. 45.6%), and more likely to be hospitalized in the Table 2 ).

A c c e p t e d M a n u s c r i p t 

Among the 29,659 patients with an initial COVID-19 related hospitalization, the overall median length of initial hospitalization was 8 days. Most of the patients were discharged home (70.1%), while 23.8% were discharged to a long-term/skilled care facility; 5.1% transferred to another hospital. Overall, 12.1% of patients required intensive care unit (ICU) care, 1.8% high-flow oxygen within 24 hours and 3.7% mechanical ventilation or ECMO during hospitalization ( Table 2) .

Patients who were readmitted were more likely to have initially been discharged to a longterm/skilled care facility compared to patients who were not readmitted (30.7% vs. 23.5%, p<0.0001).

However, patients who were readmitted had significantly shorter initial length of stay (median 7 days (range:2-54) vs. 8 days (range:2-107) days, p<0.0001), and were less likely to have required ICU level care (7.9% vs. 12.2%, p<0.0001), mechanical ventilation or ECMO at any time during the first hospitalization (4.8% vs. 8.9%, p<0.0001) ( Table 2) .

Acute conditions of COVID-19 experienced during initial hospitalization are summarized in Tables 2 and 3 . Respiratory conditions were common with 78.5% of patients experiencing pneumonia during hospitalization, 16.5% acute respiratory failure and 1.7% ARDS. Patients who were readmitted were less likely than those who were not to have been diagnosed with pneumonia during their initial hospitalization (71.1% vs. 78.7% respectively) though the rates of acute respiratory failure (17.3% vs. 16 .5%) and ARDS (1.2% vs. 1.7%) were similar between those who were and were not readmitted.

A c c e p t e d M a n u s c r i p t Several extra-pulmonary manifestations of COVID-19 were more common during the first hospitalization among patients who were readmitted vs. not readmitted, including AKI (15.6% vs. 9.2%, p<0.0001), cardiomyopathy (2.1% vs. 0.8%, p<0.0001) and congestive heart failure (6.4% vs.

2.4%, p<0.0001). Disseminated intravascular coagulation (DIC) (0.3% vs. 0.1%, p=0.03) and venous thromboembolism (2.1% vs. 1.3%, p=0.03) were significantly more common in the first admission for those who were readmitted ( Table 3) .

Risk factors associated with readmission are displayed in Figure 2 . In the multivariable 

The principal diagnosis at the time of readmission was COVID-19 in 461 (43.1%) patients.

Among these patients, pneumonia (83.9%) and acute respiratory failure (49.2%) were the most common secondary diagnoses, followed by acute heart conditions or heart failure (31.2%), AKI (27.3%) and thrombosis (11.9%) (Supplemental Table 3 ). In addition, 230 (21.5%) were readmitted with a principal diagnosis of sepsis. Of these, 29.1% had a secondary diagnosis of severe sepsis with septic shock. The most common acute conditions present on admission or during the second A c c e p t e d M a n u s c r i p t hospitalization were acute respiratory failure (30.0%, n=321), AKI (21.4%, n=229), and sepsis (17.6%, n=188) (Supplemental Table 4 ).

At the end of follow up, 49.8% of readmitted patients were discharged to home, 29.6% to a long-term/skilled care facility, 12.3% died, 3.4% were discharged to hospice, 3.2% were transferred to another hospital, and 0.2% were still hospitalized (data not shown). Dispositions at the first and second hospitalizations are summarized in Table 2 and Supplemental Table 5 . Nearly 31% of patients who were readmitted were discharged to a long-term/skilled care facility following the first admission, and 19.1% of patients were discharged to a long-term/skilled care facility following both the first and second hospitalizations.

Results from multivariable logistic regression analysis of death during readmission are displayed in 

A total of 6,819 patients were discharged <30 days prior to the study end date, including 92 who were readmitted. Sensitivity analyses excluding these patients did not alter results (data not shown).

This study includes a broad U.S. representation of hospitals and a large-scale analysis of COVID-19 readmissions. The 30-day readmission rate with 29,659 patients hospitalized for COVID-19 was 3.6%, and the rate of readmission was highest within 10 days of discharge from first York City, [15] Rhode Island, [16] and Spain, [11] suggesting that patients with these conditions who are hospitalized for COVID-19 may need extra monitoring post-discharge.

during their first admission including AKI, cardiomyopathy, and congestive heart failure were significantly higher in readmitted patients, suggesting that for patients with respiratory stability that allows for initial discharge, other organ damage must also be carefully considered.

Compared to patients who were not readmitted, a greater proportion of readmitted patients had been discharged to a long-term/skilled-care facility after first hospitalization. The high rate of readmission in patients discharged to skilled-care facilities is consistent with findings from previous studies, [8, 17] suggesting that strategies to reduce readmissions in COVID-19 patients should include careful assessment of the level of care that can be provided in a long-term care facility. Furthermore, readmitted patients had a shorter average length of stay during first hospitalization than those who were not readmitted. This is similar to patterns in other acute care hospitalizations, such as for heart failure, where decreased length of initial hospital stay may increase readmission risk. [18] In addition to having decreased average length of stays, readmitted patients in this cohort were also less likely to have received ICU level of care, which may indicate the need for longer and more escalated care A c c e p t e d M a n u s c r i p t strategies or more gradual de-escalation prior to discharge for hospitalized COVID-19 patients to prevent readmission.

The final discharge disposition of readmitted COVID-19 patients in this study was death or hospice in nearly 16% of cases, which is higher than mortality rates previously published in studies of readmission following hospitalizations for heart failure and major cancer surgery. [19, 20] On multivariable analysis, conditions noted with readmission included CKD and cardiovascular disease. Notably, CKD has been shown to predict readmissions in patients with sepsis as well, [21] and stage 5 CKD very strongly predicted readmission in this study, suggesting the need for close postdischarge monitoring of patients with this and other predictors of readmission. Patients who are insured by Medicare and Medicaid also had higher odds of readmission compared to those with commercial insurance, which may be an indication of illness, age, and/or socioeconomic status. Older age was itself found to be a risk factor for readmission. Additionally, the presence of 3 or more comorbidities was predictive of readmission, which supports the use of comorbid disease burden in risk assessment models for all-cause readmissions. [22] Overall in-hospital mortality on readmission was 12.3%. In adjusted analysis, significant predictors of inpatient mortality were oxygen therapy, sepsis and acute kidney injury. Sepsis and AKI have been notorious causes for readmission and mortality. [23, 24] This is not surprising as several immune dysfunction mechanisms may be predisposing to sepsis in debilitating conditions of cirrhosis, AKI, and heart failure, and those infected with SaRS-COv2. [25] A recent report from CDC found a higher rate of readmission than was found in the current study (~9% vs. 3.6%), possibly due to the longer window considered for readmissions of 60 days vs. 30 days, respectively, although most patients were readmitted within 10 days of discharge from their first hospitalization. This result is consistent with a study in Veterans Affairs showing patients were more likely to be readmitted within 10 days. [10] Nevertheless, both studies demonstrated that comorbid medical conditions and older age were associated with a higher likelihood of readmission A c c e p t e d M a n u s c r i p t and provide insights that may help mitigate risks for readmission. Several smaller studies have also evaluated rates of readmission in cohorts ranging from 19-328 patients. [16, 26] [27] The identification of predictors and patterns of readmission will allow for the development of targeted interventions for hospitalized COVID-19 patients. For example, in Medicaid patients with one or more chronic conditions, post-discharge engagement was associated with decreased likelihood of readmissions. [28] Given the need for social distancing measures due to the COVID-19 pandemic, telehealth platforms may be useful for post-discharge monitoring of high-risk patients. A 2017 study evaluating a telehealth protocol to prevent readmissions among high-risk congestive heart failure patients reported excellent patient adherence, indicating that discharged patients may benefit from this type of intervention. [29] The COVID-19 chargemaster data provide a substantial geographic and longitudinal distribution of inpatient data with a large sample size that allows for retrospective, longitudinal analysis of hospitalized patients, including an assessment of comorbidities, complications, and outcomes. A potential limitation of this study is that these admissions largely occurred in the early phase of the pandemic in the US, when there may have been regional differences in outcomes when hyperendemic transmission in the Northeast overwhelmed some hospital systems. Although used extensively for database research, use of ICD-10 codes also has the potential for misclassification and underrepresentation of comorbid conditions, therapies, and procedures. Data such as race, vital signs and laboratory test results were not available within the chargemaster data set.

In conclusion, in this large national cohort of patients who survived an initial admission for complications of COVID-19, readmissions demonstrate the presence of significant ongoing morbidity and mortality that may occur following initial hospital discharge and identified populations in whom additional targeted interventions are warranted to improve post-hospitalization outcomes.

A c c e p t e d M a n u s c r i p t NOTES Acknowledgements: This work was supported by Target RWE. Target RWE is responsible for the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review and approval of the manuscript; and decision to submit the manuscript for publication. Kyle Burleson provided data acquisition and validation support for this work. M a n u s c r i p t M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t 

Hospitalization and Mortality among Black Patients and White Patients with Covid-19

Presenting Characteristics, Comorbidities, and Outcomes Among 5700 Patients Hospitalized With COVID-19 in the New York City Area

Patient Characteristics and Outcomes of 11,721 Patients with COVID19 Hospitalized Across the United States

The lasting misery of coronavirus long-haulers

Is the readmission rate a valid quality indicator? A review of the evidence

Systematic review: the evidence that publishing patient care performance data improves quality of care

Impact of Hospital Population Case-Mix, Including Poverty, on Hospital All-Cause and Infection-Related 30-Day Readmission Rates

Characteristics, Outcomes, and Trends of Patients With COVID-19-Related Critical Illness at a Learning Health System in the United States

Readmission and Death After Initial Hospital Discharge Among Patients With COVID-19 in a Large Multihospital System

Hospital Readmissions of Discharged Patients with COVID-19

CMS/NCHS, ICD-10-CM Official Guidelines for Coding and Reporting. ( FY 2020

Reply to authors

A review of uses of health care utilization databases for epidemiologic research on therapeutics

Characterization of Patients Who Return to Hospital Following Discharge from Hospitalization for COVID-19

Readmissions among patients with COVID-19

Trends in length of stay and short-term outcomes among Medicare patients hospitalized for heart failure

Hospital length of stay and readmission: an early investigation

Heart failure: Same-hospital vs. different-hospital readmission outcomes

Comparison of Rates and Outcomes of Readmission to Index vs Nonindex Hospitals After Major Cancer Surgery

An analysis of errors found in renal vein sampling for plasma renin activity

The HOSPITAL score and LACE index as predictors of 30 day readmission in a retrospective study at a university-affiliated community hospital

Early Hospital Readmissions and Mortality in Patients With Decompensated Cirrhosis Enrolled in a Large National Health Insurance Administrative Database

Thirty-Day Readmissions After Hospitalization for Hypertensive Emergency

Characteristics of Hospitalized COVID-19 Patients Discharged and Experiencing Same-Hospital Readmission -United States

Characterization of Patients Who Return to Hospital Following Discharge from Hospitalization For COVID-19. medRxiv

Analysis of Risk Factors on Readmission Cases of COVID-19 in the Republic of Korea: Using Nationwide Health Claims Data

Postdischarge engagement decreased hospital readmissions in Medicaid populations

Telehealth Protocol to Prevent Readmission Among High-Risk Patients With Congestive Heart Failure

Acute MI, n (%)

Venous thromboembolism, n (%)

ARDS (acute respiratory distress syndrome), acute MI (acute myocardial infarction), DIC (disseminated intravascular coagulation), SIRS (systemic inflammatory response syndrome)

A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t