Acute cardiovascular hospitalizations and illness severity before and during the COVID‐19 pandemic

Abstract Background Cardiovascular disease (CVD) hospitalizations declined worldwide during the COVID‐19 pandemic. It is unclear how shelter‐in‐place orders affected acute CVD hospitalizations, illness severity, and outcomes. Hypothesis COVID‐19 pandemic was associated with reduced acute CVD hospitalizations (heart failure [HF], acute coronary syndrome [ACS], and stroke [CVA]), and worse HF illness severity. Methods We compared acute CVD hospitalizations at Duke University Health System before and after North Carolina's shelter‐in‐place order (January 1–March 29 vs. March 30–August 31), and used parallel comparison cohorts from 2019. We explored illness severity among admitted HF patients using ADHERE (“high risk”: >2 points) and GWTG‐HF (“>10%”: >57 points) in‐hospital mortality risk scores, as well as echocardiography‐derived parameters. Results Comparing hospitalizations during January 1–March 29 (N = 1618) vs. March 30–August 31 (N = 2501) in 2020, mean daily CVD hospitalizations decreased (18.2 vs. 16.1 per day, p = .0036), with decreased length of stay (8.4 vs. 7.5 days, p = .0081) and no change in in‐hospital mortality (4.7 vs. 5.3%, p = .41). HF hospitalizations decreased (9.0 vs. 7.7 per day, p = .0019), with higher ADHERE (“high risk”: 2.5 vs. 4.5%; p = .030), but unchanged GWTG‐HF (“>10%”: 5.3 vs. 4.6%; p = .45), risk groups. Mean LVEF was lower (39.0 vs. 37.2%, p = .034), with higher mean LV mass (262.4 vs. 276.6 g, p = .014). Conclusions CVD hospitalizations, HF illness severity, and echocardiography measures did not change between admission periods in 2019. Evaluating short‐term data, the COVID‐19 shelter‐in‐place order was associated with reductions in acute CVD hospitalizations, particularly HF, with no significant increase in in‐hospital mortality and only minor differences in HF illness severity.


| INTRODUCTION
The United States has surpassed other nations to become the epicenter of the Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) pandemic, with recent numbers of confirmed coronavirus 2019 (COVID-19) cases passing 86.6 million and 1.87 million COVID-19-related deaths in the world as of January 5, 2021. 1 Hospitals across the US and Europe have reported a significant impact on cardiovascular disease (CVD)-related hospitalizations due to the COVID-19 pandemic. [2][3][4][5][6][7][8][9] With shelter-in-place orders issued to curb the spread of SARS-CoV-2, these mandates have additionally affected the numbers of total hospital admissions 8 and raised concern for delayed access to emergent and urgent healthcare and excess CVD mortality. 2 Acute CVD conditions, including acute heart failure (HF), acute chest coronary syndromes (ACS), and acute stroke syndromes (CVA) contribute to a large disease burden and account for the vast majority of cardiovascular-related hospitalizations and deaths in the United States. 10 There is growing concern that patients with prior CVD or

| METHODS
We conducted a retrospective study using data extracted from the  11 The study received approval from the Duke University Institutional Review Board.
This study included adult patients who were admitted to a DUHS hospital from January 1, 2019 to August 31, 2019 and January 1, 2020 to August 31, 2020 with a primary International Classification of Disease (ICD)-9 or − 10 admission diagnosis of an acute CVD condition, including acute HF, ACS, or CVA (Supplemental Table 1).
Patients less than 18 years of age at time of admission were not included. Patients who had scheduled outpatient visits within the hospital that did not result in an inpatient admission were also not included. Acute CVD conditions presenting concurrently at time of admission, while uncommon, were categorized as follows: (1)  condition to cardiovascular disease, was extracted using prior historical or admission ICD-9/−10 codes (Supplemental Table 1). Place of care was defined as the primary service that cared for the patient prior to hospital discharge. Discharge disposition was determined based on discharge documentation, and categorized as home (including with and without home-based health services), facility (acute or skilled nursing care), or any hospice-based care. Length of stay (LOS) was calculated from the date and time of admission to the date and time of discharge, to the nearest tenth decimal in days. Total and inhospital mortality were recorded for all hospitalizations.
HF-specific mortality risk categories were derived from HF admission vital signs and laboratory data using the "Acute Decompensated Heart Failure National Registry" (ADHERE) and the "Get with The Guidelines ® -Heart Failure" (GWTG-HF) algorithms. [13][14][15] An ADHERE score of >2 was considered "high risk" and correlated with a 19.8-21.9% in-hospital mortality risk from acute heart failure, 13,15 and GWTG-HF score of >57 correlated with >10% in-hospital allcause mortality risk. 14 Transthoracic echocardiography (TTE) parameters were obtained for those admitted with acute HF either during the admission encounter, or, if not available, within 1.5 years prior to admission. TTE parameters included left ventricular ejection fraction (LVEF) and left ventricular (LV) mass using the linear method equation. 16 We analyzed LVEF and LV mass since they are both prognostic clinical measures of hospitalized HF patients. [17][18][19] We summarized baseline characteristics using descriptive statistics (mean for continuous variables, frequencies and percentages for binary and class variables). We compared hospitalization characteristics between admission periods and tested for differences of each measure using two-sample Student's t-test or one-way analysis of   Table 2). Exploratory analyses demonstrated reduction in daily total CVD, HF, and ACS admissions following the first NC COVID-19 case in March 2, 2020 with 2019 as reference groups (Supplemental Table 3).
There was a significant decrease in average daily acute CVD which were not present among the 2019 admission groups (Figure 2).
Despite N-terminal pro-brain natriuretic peptide levels available for 2016 (46%) of all HF admissions, there were no differences in natriuretic peptide levels across four admission periods (Table 3).

| DISCUSSION
We present a timely analysis describing the impact of the COVID-19 pandemic and a statewide shelter-in-place order on hospitalization metrics for acute CVD, including HF, ACS, and CVA, with a detailed description of risk profiles of admitted patients with HF. There was a significant reduction in average daily CVD admissions across a large academic health system after the NC shelter-in-place order on March 30, 2020. When evaluated separately, this reduction was driven primarily by a decrease in HF hospitalizations, with a significant, though less substantial, decrease in ACS hospitalizations. Despite a common concern that patients hospitalized after shelter-in-place orders may be sicker due to delayed care, our HF cohort had only a small change in The results of CVD admission trends in our study are consistent with data that has been reported at other institutions at the outset of the COVID-19 pandemic. 2 (Table 2).
Other centers, in Mississippi, United States, 6 Tennessee, United States, 4 London, United Kingdom, 3 noted significant reductions in HF hospitalizations when examined independently.
Only one other study has characterized changes in HF hospitalizations during the COVID-19 pandemic after shelter-in-place orders were lifted.
Ling et al. described a return to prior baseline hospitalization rates within 2 weeks of lifting the shelter-in-place orders in Georgia, United States. 8 In the present study, we found HF admission rates returned to 2019 levels within 6 weeks after implementation of the NC shelter-in-place order, and within 2 weeks prior to lifting this order on May 23, 2020. 12 Notably, Ling et al. noted no evidence of a "hospitalization debt" with no significant increase in hospitalizations after shelter-in-place orders were lifted. This observed phenomenon is consistent with our findings as well.
There was concern at the outset of the COVID-19 pandemic that patients presenting to the hospital would represent a sicker popula- There are many possible reasons why HF hospitalizations decreased during the study period. Successful telemedicine implementation potentially maintained outpatient care for HF patients. 21 We observed an increase in telehealth visits for CVD after the NC shelterin-place order (data not shown), which was consistent with prior published analyses on outpatient CVD care at our own institution. 23 Providers may also have managed worsening HF symptoms in same-day access clinics or observation units to prevent hospital admissions. 24,25 Additionally, patients may have been more willing to increase oral diuretics than in the past, or they may have had fewer acute exacerba-