key: cord-1000259-uv78dzna
authors: Marsh, Kassandra; Decano, Arnold; Siegfried, Justin; Ahmed, Nabeela; Blum, Sharon; Tirmizi, Samad; Dong, Mei Qin; Mehta, Dhara; Pham, Vinh P.; Papadopoulos, John; Dubrovskaya, Yanina
title: Outcomes of COVID-19 Patients Hospitalized at Acute Care Services: Real-World Experience in the New York Metropolitan Area During the Early Pandemic Before Initiation of Clinical Trials
date: 2020-12-29
journal: Infect Dis Clin Pract (Baltim Md)
DOI: 10.1097/ipc.0000000000000982
sha: f67cab76eb20a68d220a291422889477de5cc242
doc_id: 1000259
cord_uid: uv78dzna

As New York became the epicenter of the COVID-19 pandemic early on, clinicians were challenged to provide optimal medical and pharmaceutical care, despite the paucity of supporting literature and guidance. We sought to describe prescribing patterns and outcomes of physician response to the urgent need to treat COVID-19 patients before initiation of randomized clinical trials. METHODS: This was a retrospective cohort study of adult patients with COVID-19 initially admitted to acute care services during March 2020. Critically ill patients requiring intensive care unit level of care on admission were excluded. RESULTS: A total of 639 consecutive patients (supportive care, n = 247; treatment n = 392) were included in the analysis. Overall, the 28-day mortality rate was 12.2%. The mortality was 8.7% higher in the treatment group (15.6% vs 6.9% in the supportive care group, P < 0.001). Treatment was not protective against progression to severe disease (18.4% vs 3.6% with supportive care, P < 0.0001). Time to defervescence, duration of oxygen support, and hospital and intensive care unit (ICU) length of stay were also higher in the treatment group. In multivariate analysis, 60 years or older, presence of severe disease, and need for ICU admission were identified as independent predictors of 28-day mortality. There were 41 (10.5%) adverse event in the treatment group, with the majority being QT prolongation and gastrointestinal effects. CONCLUSIONS: In this cohort of hospitalized patients admitted to acute care services, treatment with hydroxychloroquine, lopinavir/ritonavir or both could not be shown to improve mortality, progression to severe disease, or clinical response.

I n December 2019, a novel coronavirus (SARS-CoV-2) detected in central China was associated with a new pneumonia syndrome . Over a 2-month period, this infection became classified as a pandemic, resulting in substantial morbidity and mortality worldwide. As New York became the epicenter of the pandemic early on, clinicians were challenged to provide optimal medical and pharmaceutical care, despite the paucity of supporting literature and guidance. 1 Pharmacotherapeutic agents, such as hydroxychloroquine (HCQ) and lopinavir/ritonavir (L/R) have been previously studied in vitro for SARS-CoV and MERS-CoV. [2] [3] [4] [5] An early study from France reported a virologic response in 26 COVID-19 patients treated with HCQ with or without azithromycin, suggesting that HCQ may lead to faster SARS-CoV-2 clearance than supportive care or observation. However, this study reported on a very small population size, had a high rate of patients lost to follow-up, and had a primary outcome that was of unclear relevance to clinical patient outcomes. 6 Nitazoxanide has activity against MERS CoV and betacoronavirus growth in mice, and in vitro activity against SARS-CoV-2. 7 Anti-interleukin-6 medications, such as tocilizumab, have been considered in COVID-19 patients with evidence of cytokine storm. 8 Finally, zinc inhibits RNA-dependent RNA polymerase and has been shown to do this in vitro against SARS-CoV. When combined with HCQ, cellular uptake is increased, making it more likely to achieve elevated intracellular concentrations. 9 Despite the lack of published literature describing the safety and efficacy of these agents for COVID-19 in humans during the early pandemic, many of the mentioned medications were frequently used "off-label" in clinical practice. This usage has led to perceptions of efficacy that, while receiving media coverage, lacked robust evidence supporting use. Randomized clinical trials are necessary to provide support for effective treatment options during this and any future outbreaks of SARS-CoV-2 associated disease. This retrospective review describes real world experience managing the COVID-19 hospitalized population in the New York metropolitan area during the early pandemic when physician response to the urgent need for patient care took place before initiation of randomized clinical trials.

This was a retrospective cohort study of consecutive patients aged at least 18 years with COVID-19 (confirmed by positive SARS-CoV-2 reverse-transcriptase polymerase-chain-reaction [RT-PCR] test results) from March 1, 2020, to March 31, 2020, and hospitalized with symptoms consistent of COVID-19 at New York University (NYU) Langone Health (Tisch Hospital/Kimmel Pavilion, NYU Brooklyn, and NYU Winthrop campuses) and Bellevue Hospital (a New York City [NYC] Health + Hospitals facility affiliated with NYU). Patients were excluded if they required an intensive care unit (ICU) level of care on admission or at the start of HCQ or L/R therapy (whichever came first), had received HCQ or L/R as home medications for other indications or for COVID-19 within the last 30 days, were enrolled in clinical trials for sarilumab versus placebo or clazakizumab versus placebo, had taken any drug not Food and Drug Administration (FDA) approved for the treatment of COVID-19 (except azithromycin) with the intent to treat COVID-19 within 30 days before admission, received less than 48 hours of HCQ or L/R during admission, or were pregnant.

Electronic health records (EHR; Epic Systems, Verona WI) were reviewed to collect data on baseline demographics, medical history, characteristics of the disease, laboratory values, and treatment used. QTc value changes were documented based on scanned electrocardiograms in the EHR, while consideration of "QT prolongation" as an adverse event (ADE) was based on physician documentation. Medication characteristics that were collected included dose, frequency, and duration of therapy. The presence of SARS-COV-2 was determined by RT-PCR of nasopharyngeal or oropharyngeal samples collected on swabs. Initially, tests were completed by the New York City Department of Health and Mental Hygiene for NYU Langone Health Tisch Hospital/Kimmel Pavilion, NYU Brooklyn Hospital, and NYU Winthrop Hospital, and a third party commercial laboratory was utilized for Bellevue Hospital samples. After March 16, 2020, the NYU Langone Health clinical laboratory conducted tests for Tisch Hospital/Kimmel Pavilion, NYU Brooklyn Hospital and NYU Winthrop Hospital using the Roche SARS-CoV-2 assay in the Cobas 6800 instruments, and only nasopharyngeal samples were tested. After April 2, 2020, NYU Winthrop Hospital's clinical laboratory conducted tests for NYU Winthrop Hospital using the Cepheid Xpert instrument, and only nasopharyngeal samples were tested. Symptoms consistent of COVID-19 included, but were not limited to, any of the following: fever, cough, dyspnea, diarrhea, nausea, and diffuse myalgia. The World Health Organization (WHO) classification was used to determine severity of infection, with 3 to 4 considered mild disease, and 5 to 7 considered severe disease. 10 

Interim Guidance for the use of potential therapeutic treatments for COVID-19 was developed by the Departments of Medicine and Infectious Diseases at NYU Langone Health, with pharmacotherapy department review and updated regularly as additional data became available. 11, 12 Early in the development, the guidance had included use of L/R monotherapy, and HCQ was added in mid-March after data from another institution that suggested benefit was publically available. Both L/R and HCQ were unrestricted and prescribed based on the physicians' discretion, taking into consideration severe drug-drug interactions (ie, L/R with apixaban, rivaroxaban, ticagrelor, clopidogrel, tacrolimus, antiepileptics, and other medications). All patients who received L/R were tested and confirmed negative for human immunodeficiency virus (HIV) infection before starting treatment. The recommended HCQ dosing regimen was 400 mg twice daily for 2 doses followed by 200 mg twice daily for 8 doses. Lopinavir/ritonavir was dosed as 400 mg/100 mg twice daily for 7 days. Order panels with defaulted dosing regimens and duration of therapy were developed within the computerized physician order entry system. Adjunctive off-label medications utilized during this period included azithromycin (500 mg once followed by 250 mg once daily for 4 days), tocilizumab (400 mg once), zinc (220 mg twice daily for 5 days in conjunction with HCQ), nitazoxanide (500 mg twice daily for 5 days), and corticosteroids at the discretion of the primary team with support from the Infectious Disease and Critical Care Medicine consultation services.

Our primary endpoint was 28-day mortality from the date of admission and time to death. Our secondary endpoints included progression to severe disease, treatment response, and treatmentrelated ADE. Progression to severe disease was assessed by the need for ICU admission, development of hypotension requiring vasopressor support, or need for escalation of oxygen supplementation to endotracheal tube or extracorporeal membrane oxygenation. Treatment response was assessed by time to defervescence, time of supplemental oxygen requirement, and length of stay in the ICU and/or hospital. We report the outcomes for the full study cohort, patients receiving HCQ or L/R or both (treatment group), and patients who received supportive care only (supportive care group). We also compared patients who died with survivors to identify potential risk factors for mortality.

All analyses were performed using SPSS Statistics Software (IBM Corp., Armonk, NY; version 25.0). Categorical variables were described as frequencies and proportions and compared using χ 2 test or Fisher exact test. Continuous variables were described as medians with interquartile ranges (IQR) and analyzed using Mann-Whitney U test. P values of 0.05 or less denote statistical significance. We also conducted a univariate analysis to compare patients who died with survivors. Variables with P values of 0.2 or less on univariate analysis were included in the multivariate logistic regression model to control for differences between groups and identify independent risk factors associated with death. The Hosmer-Lemeshow goodness-of-fit test was used to test the power of the model. A Kaplan-Meier survival curve was performed to illustrate the probability of survival from hospital admission to day 28 in the supportive care and treatment groups.

The study was approved by the NYU Grossman School of Medicine Institutional Review Board. A waiver of informed consent was granted.

Of 935 patients with COVID-19 who were admitted to the hospital during March 2020, a total of 296 were excluded (Fig. 1) . Requirement of ICU level of care on admission or at the start of treatment with L/R or HCQ or both was the most common reason for exclusion (153/296; 51.7%). A total of 639 consecutive patients with a confirmed COVID-19 test who were initially admitted to acute care services in March 2020, and either died or were discharged before the cut off day of May 8, 2020, were included in the analysis.

Among 639 included patients, 247 received supportive management (supportive care group) whereas 392 patients received off-label COVID-19 antivirals (L/R, n = 36; HCQ, n = 329; or both, n = 27, treatment group). Patients in the treatment group who received both HCQ and L/R received them sequentially (ie, not started on the same day), however may have had some days of concomitant use. There were significant differences between the 2 groups in baseline demographic characteristics, selected vital signs, laboratory values and comorbidities (Table 1) . Of note, patients at higher risk of progressing to severe disease were more likely to be treated with off-label medications at the discretion of the primary team and are therefore more frequently in the treatment group. The median time to first dose of HCQ from the time of admission was 1 day (IQR, 1-2 days), with a median total inpatient duration of therapy of 5 days (IQR, 4-6 days). The median time to first dose of L/R from the time of admission was 2 days (IQR, 1-5 days), with a median total inpatient duration of therapy of 4 days (IQR, 3-6 days). More patients in the treatment group received concomitant azithromycin ( 6-13 days] in the supportive care group, P < 0.001). Treatment was not protective against progression to severe disease (18.4% vs 3.6% with supportive care, P < 0.001). Delayed time to defervescence, prolonged duration of oxygen requirements, and prolonged hospital and ICU lengths of stay were also more frequent in the treatment group (Table 2 ). Time to death from the date of hospital admission in the overall cohort, supportive care group, and treatment group was 10 days (IQR, 6-13 days), 4 days (IQR, 3-11 day), and 10 days (IQR, 7-14 days), respectively (P < 0.001). A Kaplan-Meier analysis (Fig. 2 ) demonstrated lower cumulative survival in the treatment group compared with supportive care group, but the difference was not statistically significant (P = 0.915).

There were a total of 41 (10.5%) ADEs in the treatment group, with the majority being QT prolongation (63%) and gastrointestinal effects (20%) ( Table 2 ). Premature discontinuation of HCQ before completion of the recommended regimen occurred in 67 (17%) patients, primarily due to physician preference and patient improvement or discharge (52%). The median baseline QTc in the treatment group was 439 ms (IQR, 424-457 ms). In the treatment group, 41 (10%) patients were receiving other QT prolonging medications, of which 12 of these patients had their QT prolonging agent held for HCQ administration. The QTc in the treatment group increased during admission by 30 ms or greater in 120 (34.9%) patients, by 60 ms or greater in 52 (15.1%) patients and from less than 500 to 500 ms or greater in 52 (15.1%) patients. Hydroxychloroquine was discontinued before completion of the recommended treatment regimen due to QT prolongation in 4 (1%) patients, none of which were on concomitant QT prolonging medications.

In univariate analysis, patients who received treatment with HCQ, L/R, or both agents; received nitazoxanide; presented with severe disease on admission; required ICU admission; had cardiovascular disease including heart failure, coronary artery disease, hypertension, or hyperlipidemia; or had chronic kidney disease were more likely to die (Table 3 ). In multivariate analysis, after controlling for differences between groups, 60 years or older (odds ratio [OR], 5.2; 95% confidence interval [CI], 2.54-10.7, P < 0.001), presence of severe disease on admission (OR, 4.5; 95% CI, 1.3-15.7, P = 0.017), and need for ICU admission during the hospital stay (OR, 11.8; 95% CI, 5.9-23.5, P < 0.001) were identified as independent predictors of 28-day mortality.

In this retrospective cohort study of consecutive patients at 4 large academic centers in the New York metropolitan area early 13, 14 Because of the prescribing patterns early in the pandemic, similarly to our population, patients treated with HCQ in both of these studies were more severely ill at baseline compared with those who were not treated, and the mortality rate in the treatment group was higher before controlling for differences in patients' characteristics or propensity matching. An early smaller retrospective analysis of patients hospitalized with COVID-19 in 16 This mortality rate is much lower compared with the 28-day mortality rate of 12.2% found in our study (8.6% in treatment vs 6.9% in supportive care group), likely explained by different definitions of mild-to-moderate disease, based on supplemental oxygen requirements vs WHO classification on admission used in our study. 10 Despite the ideal randomized design by Cavalcanti et al, 16 limitations include the long time to treatment from symptom onset (up to 14 days) and inclusion of patients who began treatment before randomization in the study. Comparatively, time to initiation of COVID-19 off-label antivirals from symptom onset was shorter in our study, reflecting real word practice before randomized trials, where the need to consent, randomize, and wait for other study logistics before initiation of therapy was not a factor in patient care. We found a 10.5% (41/392) incidence of ADE with the use of HCQ and/or L/R in our population based on documentation in the EHR, with the majority being QTc prolongation and gastrointestinal effects including nausea, diarrhea, and gastroesophageal reflux. In a retrospective analysis of 1061 cases, an overall rate of ADE with HCQ and azithromycin was reportedly lower at 2.4% (25/1061), consisting mainly of gastrointestinal symptoms (18/25, 72%), with only 3 patients requiring discontinuation of treatment because of ADE. 17 In a randomized, double-blind, placebo-controlled trial of 423 patients with COVID-19, medication ADE occurred in 43% (92/212) of participants receiving HCQ versus 22% (46/211) receiving placebo (P < 0.001). 18 Of the HCQ-related ADE, gastrointestinal symptoms were the most commonly reported, with 31% (66/212) reporting upset stomach or nausea, 24% (50/212) reporting abdominal pain, diarrhea, or vomiting, and no patients experiencing cardiotoxicity. 18 There were several limitations of this review. Because of the retrospective, observational study design, there were significant baseline differences between the supportive care and treatment groups, which may reflect a willingness to give patients at higher risk for progression of disease during the early pandemic an unproven treatment in the absence of proven options. Despite use of a binary logistic regression model to adjust for these differences, we cannot exclude the possibility of unmeasured confounders. Collection of ADE and QT prolongation relied on documentation in the EHR and may, therefore, underestimate the true incidence. Our description of QTc changes relied on uploaded electrocardiograms into the EHR and did not include data being monitored via telemetry without EHR documentation. In addition, evaluation of 28-day mortality has some limitations because of the retrospective design, including the inability to capture mortality in the outpatient setting if not updated in the EHR. The primary outcome of in-hospital mortality was expanded to 28-day mortality because of the knowledge of COVID-19-related readmissions leading to an outcome of death within 14 days of the index admission. We also were unable to assess virologic effects because of the absence of standardized retesting of patients and exclusion of collection of follow-up RT-PCR data. Finally, although patients from 4 large academic medical centers were included, outcomes of our patient cohort early in the pandemic still might not be generalizable to other populations.

Throughout the pandemic, health care providers may have become more knowledgeable and skilled in performing standard supportive management and care of COVID-19 patients, leading to improved outcomes as the pandemic went on. Of importance, the growing body of evidence led to the FDA Emergency Use Authorization for remdesivir and convalescent plasma for the treatment of hospitalized patients with COVID-19. Results of the RECOVERY trial provided support for dexamethasone use for the treatment of COVID-19 in hospitalized patients who are mechanically ventilated, and in hospitalized patients who require supplemental oxygen but who are not mechanically ventilated. 19 Based on published data, Infectious Diseases Society of America and National Institute of Health (NIH) COVID-19 guidelines recommend against the use of HCQ for the treatment of COVID-19 in hospitalized patients. 11, 12 Among patients who have been admitted to the hospital with COVID-19, the Infectious Diseases Society of America guideline panel recommends the combination of L/R only in the context of a clinical trial. 12 Our local COVID-19 Guidance has been updated on a regular basis throughout the pandemic to account for these recommendations. Overall, this retrospective observational study adds to the collective knowledge of real-world practice during the early COVID-19 pandemic by describing prescribing patterns and evaluating the mortality and safety outcomes in hospitalized patients initially admitted to acute care services before initiation of randomized controlled trials.

COVID-19 Clinical Trial Oversight at a Major Academic Medical Center: Approach of the Michigan Medicine COVID-19 Clinical Trial Committees

Hydroxychloroquine, a less toxic derivative of chloroquine, is effective in inhibiting SARS-CoV-2 infection in vitro

Role of lopinavir/ritonavir in the treatment of SARS: initial virological and clinical findings

Treatment with lopinavir/ritonavir or interferon-β1b improves outcome of MERS-CoV infection in a nonhuman primate model of common marmoset

Comparative therapeutic efficacy of remdesivir and combination lopinavir, ritonavir, and interferon beta against MERS-CoV

Hydroxychloroquine and azithromycin as a treatment of COVID-19: results of an open-label non-randomized clinical trial

A screen of the NIH clinical collection small molecule library identifies potential anti-coronavirus drugs

Effective treatment of severe COVID-19 patients with tocilizumab

Zn(2+) inhibits coronavirus and arterivirus RNA polymerase activity in vitro and zinc ionophores block the replication of these viruses in cell culture

COVID-19) Treatment Guidelines. National Institutes of Health

Infectious Diseases Society of America Guidelines on the treatment and management of patients with COVID-19

Observational study of hydroxychloroquine in hospitalized patients with Covid-19

Association of treatment with hydroxychloroquine or azithromycin with in-hospital mortality in patients with COVID-19 in New York State

Outcomes of hydroxychloroquine usage in United States veterans hospitalized with Covid-19

Coalition Covid-19 Brazil I Investigators. Hydroxychloroquine with or without azithromycin in mild-to-moderate Covid-19

Early treatment of COVID-19 patients with hydroxychloroquine and azithromycin: a retrospective analysis of 1061 cases in Marseille, France

Hydroxychloroquine in nonhospitalized adults with early COVID-19: a randomized trial

Dexamethasone in hospitalized patients with Covid-19-preliminary report