key: cord-0966890-2664hmz9
authors: Cottini, Marcello; Lombardi, Carlo; Berti, Alvise; Gregis, Marco; Gregis, Giorgio; Bello, Luigi; Mazid, Mahmoud; Putignano, Tommaso; Corbellini, Aldo; Belotti, Stefano; Rossi, Sergio; Finazzi, Alfredo; Locatelli, Mario; Zelaschi, Fabrizio; Raimondo, Maria; Miscia, Rossana; Ferrari, Fabio; Chiodini, Annalisa; Rovelli, Monica; Locatelli, Cecilia; Narzisi, Iolanda; Staats, Jutte; Mazzoleni, Leonello; Breviario, Adele; Sequenzia, Federica; Scorpiniti, Anna; Barili, Daniela; Cocchiola, Margherita; Donatini, Rinaldo; Invernici, Rosalba; Sergio, Francesco; Munizza, Titti; Travella, Bruno; Luderin, Lucilla; Gotti, Rossella; Rampinelli, PierLuigi; Roberto, Bettini; Locatelli, Giuseppe
title: Obesity is a Major Risk Factor for Hospitalization in Community-managed COVID-19 Pneumonia
date: 2021-02-04
journal: Mayo Clin Proc
DOI: 10.1016/j.mayocp.2021.01.021
sha: ae86c25df2de41616049a9a06cef8dc18e3bed32
doc_id: 966890
cord_uid: 2664hmz9

Objective We aimed to investigate whether the stratification of outpatients with Coronavirus disease 2019 (COVID-19) pneumonia by body-mass index (BMI) can help to predict hospitalization and other severe outcomes. Patients and Methods We prospectively collected consecutive cases of community-managed COVID-19 pneumonia from March 1st to April 20th 2020 in the province of Bergamo, and evaluated the association of overweight (25kg/m2≤BMI<30kg/m2) and obesity (≥30kg/m2) with time-to-hospitalization (primary endpoint), low-flow domiciliary O2 need, non-invasive mechanical ventilation (NIV), intubation and death due to COVID-19 (secondary endpoints) in this cohort. We analyzed the primary endpoint using multivariable Cox models. Results Of 338 patients included, 133 (39.35%) were overweight and 77 (22.78%) were obese. Age at diagnosis was lower in obese patients compared to those with over- or normal weight (p<.001), while diabetes, dyslipidemia and heart diseases were differently distributed among BMI categories. Azithromycin, hydroxychloroquine, and prednisolone use was similar between BMI categories (p>.05). Overall, 105 (31.07%) patients were hospitalized, and time-to-hospitalization was significantly shorter for obese versus over- or normal-weight patients (p<0.001). In the final multivariable analysis, obese patients were more likely to require hospitalization than non-obese patients (HR, 5.83; 95%CI, 3.91 to 8.71). Results were similar in multiple sensitivity analyses. Low-flow domiciliary O2 need, hospitalization with NIV, intubation and death were significantly associated with obesity (p<.001). Conclusion In patients with community-managed COVID-19 pneumonia, obesity is associated with a higher hospitalization risk and overall worse outcomes than non-obese patients.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a novel Coronavirus first detected in Wuhan, China, in December 2019, spread all over the world in the following weeks. 1 Patients infected by SARS-CoV-2 develop coronavirus disease 2019 (COVID-19), characterized by a high rate of hospitalization, respiratory failure, and ultimately death. 2, 3, 4, 5, 6, 7 Coexisting conditions, such as diabetes, hypertension, malignancy, chronic obstructive pulmonary disease (COPD), and older age are risk factors for severe disease and poor outcome in hospitalized patients. Obesity has been shown to be highly prevalent in patients requiring hospital care or invasive mechanical ventilation, [8] [9] [10] [11] 12, 13 and to be associated with in-hospital mortality after adjustment for other comorbidities. 14 In the most affected countries, the asymptomatic and non-critical patients with active infection were treated at home in order to avoid the collapse of the health systems. The Lombardy region, in Italy, and specifically the province of Bergamo, has been severely affected by COVID-19, after the first case detected in Alzano Lombardo on February 23 rd . 15 The highest rates of infection and death in Italy were registered in this province, unfortunately making this area the ideal epidemiological setting to study COVID-19.

Herein, we aimed to examine the association of increased body mass index (BMI) with hospitalization and severe outcomes in a non-hospital setting of COVID-19 pneumonia. We collected all the consecutive cases of community-managed COVID-19 pneumonia during the early weeks of pandemic (March 1 st to April 20 th , 2020) from a large cohort of residents in province of Bergamo followed by 35 primary care providers, and evaluated the association of overweight and obesity with time-to-hospitalization (primary endpoint), domiciliary low-flow O 2 need, non-invasive mechanical ventilation (NIV), intubation and death due to COVID-19 (secondary endpoints).

We conducted a prospective observational cohort study in the province of Bergamo area, Italy.

Study participants were recruited from the adult general population (≥18 years old) among approximately 40.000 residents followed-up by 35 primary care providers (i.e. up to 1500 patients for each general practitioner, ranging from 1000 to 1500), from March 1 st 2020 to April 20 th 2020. Followup continued until death or May 31 st , 2020. The electronic medical records of the recruited outpatients were accessed by the respective providers and data were manually abstracted. The availability of this comprehensive medical data, which systematically collects all the clinical records of each individual in the province, allowed a detailed case ascertainment.

Incident cases of COVID-19 pneumonia were collected during the study period, i.e. the early phases of pandemic in Lombardy. All the adults with at least 2 out of the 4 following symptoms (fever ≥ 37.5°C, cough, pleuritic chest pain, and dyspnea) with pneumonia documented by Computer Tomography (CT) scan between March 1 st 2020 and April 20 th 2020 were included in the study.

Patients were tested by reverse-transcriptase-polymerase-chain-reaction (RT-PCR) assay for SARS-CoV-2 when clinically indicated by the local health authority. Patients not fulfilling these criteria, including asymptomatic or low-symptomatic cases without pneumonia, even if positive for SARS-CoV-2 at RT-PCR, were excluded. Patients with acute respiratory distress syndrome (ARDS) 16 and/or requiring hospitalization and/or respiratory support at onset were not included. Patients with evidence of bacterial pneumonia (i.e. clear imaging signs of bacterial pneumonia according to the radiological report) were also excluded.

Patients were treated for COVID-19 according to medical judgment, following a shared protocol provided by the referral hospital Giovanni XXIII of Bergamo, which included hydroxychloroquine (HCQ) 200 mg 12 hours apart for the first 2 doses, then 200 mg/day for ≥ 5 days; oral azithromycin 500 mg/day for ≥ 5 days; oral cefixime: 400 mg/day for ≥ 5 days; oral prednisolone or equivalents: 5-25 mg/day for ≥ 5 days; subcutaneous enoxaparin: 4000U/day until mobilization or resolution of phlebitis. In general, patients started with azithromycin with or without HCQ, and cefixime was added after 5 days if no improvement was seen, in case of macrolide allergy or in addition to previous treatments in patients with age ≥65 or ≥1 comorbidities. Prednisolone and Enoxaparin were added according to clinical judgment. General practitioners were allowed to prescribe domiciliary low-flow O 2 therapy to patients with oxygen saturation <93% at resting in ambient air documented by pulse oximeter (<90% for patient affected by COPD) or heart rate >22 beats per minute. Data on patients' demographic, baseline comorbidities, presenting symptoms, oxygen saturation in ambient air at presentation, historical and current medication list, low-flow O 2 prescription by the general practitioners, inpatient hospitalization, invasive and non-invasive ventilator use data, and death were collected.

The study was conducted according to STROBE guidelines (STrengthening the Reporting of Observational Studies in Epidemiology) for cohort studies. This study was conducted in compliance with the Good Clinical Practice protocol and the Declaration of Helsinki principles and was approved by the local institutional review board.

J o u r n a l P r e -p r o o f

The most recent patient weight and height, during the 12 months preceding the index date (diagnosis of pneumonia) were collected, and BMI was calculated as weight in kilograms divided by the square of the height in meters. Patients were stratified by BMI as: normal-weight (BMI<25kg/m 2 ), overweight (25kg/m 2 ≤ BMI <30kg/m 2 ), and obese (≥30kg/m 2 ), according to the World Health Organization definitions. 17

The primary endpoint was the time from index date to hospitalization due to COVID-19 (timeto-event outcome). Secondary endpoints were time from index date to death due to COVID-19, and the associations of BMI categories with hospitalization, domiciliary low-flow O 2 need, NIV, intubation, and death during the observation period.

Categorical data were summarized as percentages, significant differences or associations of BMI categories with secondary endpoints or other clinical features were analyzed using the X 2 test or Fisher exact tests, where appropriate. Continuous variables were presented as mean±standard deviation (SD) or median and interquartile range (IQR), depending on normality demonstrated by Kolmogorov-Smirnov test. Comparisons were performed with either Student's t-test for independent samples (2tailed) or with analysis of variance (ANOVA) comparisons with Bonferroni correction when the means of the three BMI categories were being compared.

Cox-proportional hazard regression models were used to estimate the association between obesity and hospitalization (primary endpoint). Patients without a primary endpoint event had their data censored on May 31 st , 2020. An initial multivariable Cox regression model included as covariates demographic factors, comorbidities at diagnosis, and treatment for community-managed COVID-19 pneumonia. A Cox regression model including as covariates only those with significant p values at univariate analysis was performed and reported. The estimated distribution of hospitalization and death was performed with the Kaplan-Meier method and the long-rank test. Multiple imputation was used to handle missing data, and model estimates and standard errors were calculated with Rubin's rules. 18 All the analyses were performed using JMP Pro package (SAS Institute Inc., Cary, North Carolina) and SAS System for Windows, version 9.4 (SAS Institute), and a p-value of <.05 was considered statistically significant for all the analysis.

Of the 341 consecutive patients, 3 were excluded because patients did not meet the study criteria (i.e. evidence of bacterial pneumonia on CT-scan). A total of 338 patients was included in the analysis. Distribution of BMI ranged from 16.95 kg/m 2 to 41.43kg/m 2 (Supplementary Figure 1) ; 133 (39.35%) patients were overweight and 77 (22.78%) patients were obese.

Baseline demographic and clinical features are described in Table 1 . Age at diagnosis was lower in obese (BMI≥30kg/m 2 ) patients compared to overweight (25kg/m 2 ≤BMI<30kg/m 2 ) and normal-weight (BMI<25kg/m 2 ) patients (p<.001 in both cases by direct comparison), while prevalence of male sex increased with higher BMI (p<.001 for both obese and overweight vs. normal-weight patients, by direct comparisons). Diabetes, dyslipidemia and heart disease were differently distributed among the BMI categories. By direct comparison, diabetes was more frequent in normal-weight compared to overweight patients, while dyslipidemia was more frequent in overweight compared to normal-weight patients (p<.001 for both). Heart diseases were higher in obese compared to overweight patients and in normal-weight compared to overweight patients (p=.001 and .02, respectively).

The proportions of patients reporting headache, syncope/presyncope, dyspnea at resting and oxygen desaturation (<93% in ambient air) progressively increased with the increasing of BMI categories, while rhinorrhea/nasal obstruction progressively decreased ( Table 2) . Overall, approximately 42% of patients underwent nasal swab for RT-PCT confirmation, and all of them resulted positive.

The cumulative frequency of each drug used during the observation period for the treatment of COVID-19 pneumonia is reported in Table 2 . Virtually all the patients used azithromycin, more than 50% used HCQ and approximately one third used low-medium dose of oral prednisolone or equivalents. Overall, no difference was observed between the BMI categories, with the exception of paracetamol. p<.001 for both comparisons). Time-to-hospitalization was significantly shorter for obese compared to both normal-and over-weight patients ( Figure 1B) .

In the crude, unadjusted analysis, obese patients were more likely to require hospitalization than non-obese patients (BMI<30 kg/m 2 ; HR, 6.21; 95% CI, 4.20 to 9.18). Male patients carried also higher risk (HR, 1.53; 95% CI, 1.01 to 2.31), while HCQ and prednisone use (HR, 0.68; 95% CI, 0.17, 1.00; and HR, 0.59; 95% CI, 0.38 to 0.92, respectively) were less likely associated to primary endpoint in the unadjusted analyses. In the final multivariable analysis, obese patients were more likely to necessitate hospitalization than non-obese (HR, 5.83; 95% CI, 3.91 to 8.71), without significant associations of other covariates (Figure 2) . Results were similar in multiple sensitivity analyses (Supplementary Table) . pneumonia by BMI categories, showing that obese patients were more likely to need hospitalization than non-obese patients. Moreover, obese patients had a more severe course, requiring domiciliary lowflow O 2 , NIV, and intubation. Death due to COVID-19 was also higher in obese compared to nonobese patients.

Obesity is a recognized independent predictor of severe H1N1 infection, 19,20 and a higher BMI has been inconsistently reported as a potential risk factor for severe outcomes of hospitalized COVID-19 patients. 1, 3, [8] [9] [10] [11] As for hospital cohorts, 12,13 the prevalence of obesity and overweight was high in community-managed COVID-19 patients, and was associated with worst prognoses. Interestingly, overweight/obesity and altered liver function were associated to the probability of prolonged hospitalization in patients with COVID-19 infection, highlighting the role of high BMI as a predictor of worst prognosis in hospitalized patients. 21 In particular, in our cohort obesity was associated with higher hospitalization risk and various degrees of respiratory impairment, as shown by the higher need of domiciliary low-flow O 2 , NIV and intubation during the observation period. This is consistent with several previous reports on hospitalized patients, showing the association of obesity with invasive mechanical ventilation, 12, 13 , in-hospital mortality 14 , and overall an increased risk of critical illness during the disease course, as shown from a recent meta-analysis 35 . Interestingly, higher BMI was associated with lower age at diagnosis. This was consistent with a previous observation from John

Hopkins' hospitalized patients, showing that obesity can shift severe COVID-19 disease to younger age. 22 In community-managed patients, common comorbidities such as diabetes, hypertension, renal impairment, cardiac and pulmonary diseases, which are recognized risk factors for severe COVID-19,

were not significantly associated with hospitalization. Consistently, obesity was the only variable J o u r n a l P r e -p r o o f associated with hospitalization in the final Cox regression analysis. Although it is still possible that some amount of unmeasured confounding remains, the correction of the analysis for the most significant confounders and the consistency of the results across sensitivity analyses is reassuring.

Since the primary outcome measure of this study (i.e. hospitalization) is different from those of the previous hospital-cohort studies (mostly intubation and/or in-hospital death), our findings suggest that pre-existing patient conditions driving hospitalization risk are likely different from those of poor outcomes in hospitalized patients.

This role of obesity in predisposing worst outcomes in initially non-critical COVID-19 pneumonia, leads to several practical considerations. First, for the clinicians and patients, active vigilance and more aggressive management should be recommended for obese patients with COVID-19 pneumonia treated in a non-hospital setting. Second, since patients with higher BMI do not only have higher hospitalization risk, but also a higher need of respiratory support, patients with BMI higher than 30 kg/m 2 might be preemptively hospitalized regardless their clinical condition, in the attempt to promptly intervene in case of respiratory or general complications. In non-COVID19 clinical studies, obesity has been recognized as an independent predictor of ARDS in mechanically ventilated critically-ill patients. 23 Several studies depicted obesity as a chronic low-grade inflammatory condition; 24,25 others showed an effect on pulmonary function, impacting on lung volumes and compliance, and narrowing the peripheral airways. 26,27,28,29 Since ACE2 is highly expressed in adipose tissue, more than in the lungs, it has been hypothesized that SARS-CoV-2 could be able to enter into adipocytes, which could become infected thus contributing to the spreading to other organs, or represent a natural reservoir for the virus thus leading to a prolonged viral clearence. 36 In addition, obesity may induce alteration in the rennin-angiotensin-system, promoting further derangement in COVID-19. 36 Ultimately, several factors could contribute to explain the pathophysiology underlying this association, and further investigations are needed to clarify the link between obesity and severe COVID-19.

Patients with mild or moderate COVID-19 are usually managed with supportive care at home. 30 Community-managed COVID-19 pneumonia, i.e. patient with a moderate illness with clinical and radiological signs of pneumonia, relies on acetaminophen, nonsteroidal anti-inflammatory drugs (NSAIDs), azithromycin, HCQ, enoxaparin, and prednisone. No antiviral treatment or anti-cytokine monoclonal antibodies are used in this setting. No substantial difference has been observed in COVID-19 treatment across BMI categories, and in the multivariate analysis, no significant association between HCQ use and hospitalization has been observed. This result is in line with previous analyses on large cohorts of hospitalized patients, showing no protective effects of HCQ for intubation and/or in-hospital mortality, alone or in combination with azithromycin, 7,31 and with a recent randomized control trial showing that HCQ prophylaxis did not prevent illness after high-to-moderate risk exposure to SARS-

Our data are subjected to several limitations. First, the observational retrospective design and the extraction of data from non-standardized medical records cannot completely exclude classification error and survivor treatment selection. However, given the direct involvement of general practitioners in collecting the data of their own patients and the relatively limited time of observation, we believe that these potential biases are minimal, if even present. The relatively limited availability of confirmatory testing due to the Italian Government restriction for RT-PCR in non-hospitalized patients could have led to ascertainment bias. However, all our patients were diagnosed by CT scan, which has a sensitivity if 97% for detection of COVID-19 pneumonia and has been proposed as primary tool for COVID-19 detection in epidemic areas. 33 In addition, all the tested patients (42%) resulted positive to RT-PCR for Sars-CoV-2, and ultimately, it is quite accepted the inclusion of patients with indicative J o u r n a l P r e -p r o o f clinical pictures, yet not routinely tested for influenza and H1N1 in retrospective studies. 19 Finally, the results of this study may not reflect the risk factors for hospitalization in COVID-19 patients without pneumonia and/or other communities with a different ethnic composition, as those with high African Americans or Asians. 34

In conclusion, we showed that obese patients with community-managed COVID-19 pneumonia were more likely to require hospitalization and an overall more severe course than non-obese patients.

Obesity was the only factor among the pre-existing comorbidities and treatments for SARS-CoV-2 that associated with subsequent hospitalization in this community-managed cohort. Our findings should raise the level of awareness in clinicians and patients on the hospitalization risk in obese patients with COVID-19 pneumonia.

This article is dedicated to all the Italian doctors and health personnel that sacrificed their own lives to save patient lives. We wish to honor their competence, braveness and generosity. 

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Clinical characteristics of coronavirus disease 2019 in China

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Case-Fatality Rate and Characteristics of Patients Dying in Relation to COVID-19 in Italy

Baseline Characteristics and Outcomes of 1591

Patients Infected with SARS-CoV-2 Admitted to ICUs of the Lombardy Region, Italy

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

Observational Study of Hydroxychloroquine in Hospitalized Patients with Covid-19

Patients with Covid-19 in China: A Nationwide Analysis

Clinical characteristics of 140 patients infected with SARS-CoV-2 in Wuhan, China

Predictors of mortality for patients with COVID-19 pneumonia caused by SARSCoV-2: A prospective cohort study

Preliminary estimates of the prevalence of selected underlying health conditions among patients with coronavirus disease 2019 -United States

Factors associated with hospitalization and critical illness among 4,103 patients with COVID-19 disease in New York City. medRxiv

High prevalence of obesity in severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) requiring invasive mechanical ventilation