Characteristics and outcomes of COVID-19 patients with COPD from the United States, South Korea, and Europe

Background: Characterization studies of COVID-19 patients with chronic obstructive pulmonary disease (COPD) are limited in size and scope. The aim of the study is to provide a large-scale characterization of COVID-19 patients with COPD. Methods: We included thirteen databases contributing data from January-June 2020 from North America (US), Europe and Asia. We defined two cohorts of patients with COVID-19 namely a ‘diagnosed’ and ‘hospitalized’ cohort. We followed patients from COVID-19 index date to 30 days or death. We performed descriptive analysis and reported the frequency of characteristics and outcomes among COPD patients with COVID-19. Results: The study included 934,778 patients in the diagnosed COVID-19 cohort and 177,201 in the hospitalized COVID-19 cohort. Observed COPD prevalence in the diagnosed cohort ranged from 3.8% (95%CI 3.5-4.1%) in French data to 22.7% (95%CI 22.4-23.0) in US data, and from 1.9% (95%CI 1.6-2.2) in South Korean to 44.0% (95%CI 43.1-45.0) in US data, in the hospitalized cohorts. COPD patients in the hospitalized cohort had greater comorbidity than those in the diagnosed cohort, including hypertension, heart disease, diabetes and obesity. Mortality was higher in COPD patients in the hospitalized cohort and ranged from 7.6% (95%CI 6.9-8.4) to 32.2% (95%CI 28.0-36.7) across databases. ARDS, acute renal failure, cardiac arrhythmia and sepsis were the most common outcomes among hospitalized COPD patients. Conclusion: COPD patients with COVID-19 have high levels of COVID-19-associated comorbidities and poor COVID-19 outcomes. Further research is required to identify patients with COPD at high risk of worse outcomes.

Any reports and responses or comments on the article can be found at the end of the article.

Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has infected over 200 million patients and resulted in more than 4.2 million deaths worldwide as of April 2021 1 . Coronavirus disease 2019 (COVID- 19) can lead to severe lung injury and pneumonia, acute kidney injury, cardiovascular complications, and death. The symptoms and complications of COVID-19 early in the pandemic have been compared to seasonal influenza resulting in national policy measures classifying chronic obstructive pulmonary disease (COPD) patients as high risk and advising them to take additional protective measures 2 . The prevalence of COPD among COVID-19 patients has ranged from 0.8% to 38% during the first wave of the pandemic depending upon the cohort studied 3 . Whilst some studies suggested that the prevalence of COPD among COVID-19 patients during the first wave of the pandemic may be lower than the prevalence of COPD in the general population COPD was still considered a risk factor for severe COVID-19 disease 4 .
Estimates for the prevalence of COPD among COVID-19 patients from the first wave of the pandemic typically came from small, single-centre hospitalised cohorts and examined a limited range of patient characteristics and outcomes 3 . Larger comparisons from geographically diverse cohorts that also include patients with milder COVID-19 illness provide a more compelling picture and improve generalisability. Viral respiratory tract infections are common triggers for exacerbations resulting in increased morbidity and mortality yet it is uncertain how often people with COPD with COVID-19 present with exacerbations 5,6 .
The aim of this study was to perform a large-scale, federated network, descriptive characterization study reporting the demographics, comorbidities, and outcomes of COPD patients with COVID-19 during the first wave of the pandemic at the point of diagnosis and hospitalisation.

Ethical approval
All the data partners received Institutional Review Board

Study design
The Characterizing Health Associated Risks and Your Baseline Disease In SARS-COV-2 (CHARYBDIS) study is a multinational cohort study using retrospective electronic health records and claims data on COVID-19 patients from three continents during the first wave of the pandemic, the North America (US), Europe, and Asia 7 . All data for were standardized to the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) 8,9 . The Charybdis protocol and source code is available via open access (https://github.com/ohdsi-studies/Covid19Charac-terizationCharybdis) 10 .

Data sources
Of the nineteen databases available on 28 th November 2020, 13 that had a minimum sample size of 140 COVID-19 patients with COPD were included. This minimum threshold was considered appropriate to estimate the prevalence of a previous condition or 30-day risk of an outcome affecting 10% of the study population. Supplementary Figure S1 presents the database selection process for this study 11 11 .
Study participants and follow-up COVID-19 cohorts: Two non-mutually exclusive cohorts were defined (Appendix 2) 11 . COVID-19 patients in the diagnosed cohort were defined as patients during the first wave of the pandemic having a clinical diagnosis and/or positive test for SARS-CoV-2 from outpatient or inpatient records. In the diagnosed cohort, the index date was the earliest date of COVID-19

Amendments from Version 2
Sentences furthering limitations of this study have been added. These include comments on definition of the hospitalized cohort and on misclassification of asthmatics. We have also removed the lines related to hydroxychloroquine and azithromycin. Finally, we mention through all the text that this study relates to the first wave of the pandemic.
Any further responses from the reviewers can be found at the end of the article REVISED diagnosis or a first positive test. COVID-19 patients in the hospitalized cohort were defined as patients during the first wave of the pandemic with a hospitalization episode and a clinical diagnosis of COVID-19 or positive SARS-CoV-2 test within 21 days prior to admission and up to the end of hospitalization. This time window was chosen to include patients with a diagnosis prior to hospitalization and to allow for delays in recording of test results. In the hospitalized cohort, the index date was the day of hospitalization.
All patients were required to have at least a year of observation time prior to index date. Patients were followed from the index date to the earliest of the studied outcome, end of follow-up (30 days after index date), end of data capture, or death.
COPD definition: COPD was defined as either: a) an occurrence of a COPD diagnosis code any time on or before the COVID-19 index date or b) a prescription or administration of COPD medications within the year prior to index date in patients older than or equal to 55 years (Appendix 3) 11 . We excluded patients with a diagnosis of asthma prior to the COPD diagnosis to avoid misclassification with asthma. Codes used to define these cohorts have been previously described 12,13 and are included in the Appendix 11 .

Baseline characteristics
Conditions and procedures were identified within 1 to 365 to days prior to the index date using Systematized Nomenclature of Medicine (SNOMED CT) codes with all descendent codes included codes mapped from local source vocabularies. We report pre-specified demographics and conditions related to COPD and COVID-19. Other conditions analysed as part of the larger CHARYBDIS project are reported here (https://data.ohdsi. org/Covid19CharacterizationCharybdis/). COPD exacerbation was defined by a COPD exacerbation code at index date in databases with disease codes for exacerbation. The following medications were identified within 1 to 30 days prior to index date to characterise how patients were recently managed prior to the COVID-19 index date: systemic corticosteroids, inhaled corticosteroids (ICS), short-acting beta2-agonists (SABA), long-acting beta2-agonists (LABA), short-acting muscarinic antagonists (SAMA), long-acting antimuscarinic antagonists (LAMA), methylxanthines, mucolytics, oxygen therapy, antibiotics (beta-lactam penicillins, macrolides, fluoroquinolones), acetaminophen, nonsteroidal anti-inflammatory drugs (NSAIDs), and opioids. Medication use was calculated using drug eras that began starting on the date of the first drug exposure and ended on the observed end date of the last concatenated medication record, with a grace period of 30 days between medication records which allowed for sequential medication records to be considered as a continuous drug era.

Outcomes
We identified the following outcomes within 30 days following the index date: death, use of intensive services (identified as a recorded invasive mechanical ventilation and/or a tracheostomy and/or extracorporeal membrane oxygenation procedure), acute respiratory distress syndrome (ARDS), acute renal failure syndrome (ARFS), cardiac arrhythmia, heart failure, pulmonary oedema, myocardial infarction, sepsis, bleeding, venous thromboembolism (VTE), pulmonary embolism (PE) and stroke (ischaemic and haemorrhagic).

Analysis
A common analytical code for CHARYBDIS was run locally in each database. Only aggregate results from each database were then shared. We report the number and proportion by sociodemographics, history of comorbidities, and commonly used medications in each population with 95% confidence intervals (CI) calculated using the Wilson score method. Standardised mean differences (SMD) were calculated to aid comparison between study cohorts. We used R version 3.6.0 for data visualization.
All the data partners obtained Institutional Review Board (IRB) approval or exemption to conduct this study, as required.

Demographics
In the hospitalized cohort, COPD patients with COVID-19 were more commonly male (range 46.8% to 58.5%, overall median 54.5%) (Supplementary Table S1) 11 . However, there was less consistent sex difference amongst patients in the diagnosed cohort (Supplementary Table S2) 11 . Whilst in VA-OMOP 96.7% of hospitalized patients and 94.6% of diagnosed patients were male, this was expected due to the population demographics with data predominantly originating from male veterans. In both cohorts, COPD patients with COVID-19 had a similar distribution of age and were more commonly older than 65 years (Supplementary Figure S2 and S3) 11 .

Discussion
COPD prevalence among patients with COVID-19 from the first wave of the pandemic was 1.5-to 3-fold greater among those hospitalized than among those in the diagnosed cohort. Studies from China at the time reported the lowest observed COPD prevalence in COVID-19 patients with rates as low of 0.8% 3 . In contrast, COPD prevalence from the first wave appeared greater in European and US COVID-19 study populations. We similarly observed a low COPD prevalence among COVID-19 patients from the first wave in South Korea compared to other countries, which may reflect differences in the baseline prevalence of COPD in each country. However, it could also relate to differences in how health care systems responded to the pandemic, for example whether people with COPD were considered high risk and given advice on risk reduction measures and shielding. The differential response between countries during the first wave of the pandemic could therefore have influenced the prevalence measures.
COVID-19 patients with COPD from the first wave of the pandemic in both cohorts had a similar age distribution with the proportion of men being consistently higher in the hospitalized cohort. Increasing age, male gender and a history of cardiovascular disease are established risk factors for severe COVID-19 [14][15][16][17][18][19][20] . We similarly observed a high prevalence of cardiometabolic comorbidities among COPD patients with COVID-19 from both cohorts. This included arrythmia, which may be related to atrial fibrillation being prevalent within patients with COPD.     Although exacerbation at presentation was recorded more commonly among hospitalized patients, overall exacerbation prevalence among COVID-19 patients with COPD during the first wave of the pandemic was relatively low. Further studies are required to formally assess to what degree typical exacerbations of COPD are a presenting feature of COVID-19 in people with COPD.
There have been safety concerns over the role of ICS with reports of worse COVID-19 outcomes associated with ICS use 21,22 . Whilst our study was not designed to formally assess this, we saw no large differences in ICS use between the cohorts as might be expected if use was associated with a large risk. Indeed, early clinical trials suggest that use of inhaled budesonide use may be beneficial 23 . The increased use of acetaminophen, opioids and NSAIDs among hospitalized patients also suggests greater symptomatic illness. Whilst similar safety concerns with NSAID use also emerged early in the pandemic, more recent studies have not found them to be harmful 24,25 .
The most common 30-day outcomes were ARDS, ARFS, arrhythmia, sepsis and heart failure suggesting that a multi-organ approach was required for COVID-19 clinical management during the first wave of the pandemic. As expected, hospitalized COPD patients had a higher prevalence of poor health outcomes. However, it is useful to understand this risk among a cohort that includes milder cases at an earlier stage of the illness despite having similar levels of baseline comorbidity.

Strength and limitations
A strength of this study is the federated analysis allowing large numbers of patients to be characterized between countries, which overcomes some of the limitations of smaller single centre studies and inevitable heterogeneity that can occur by applying different study designs and methods of analysis. Furthermore, information on a large number of additional patient characteristics relating to conditions and treatment are also available online beyond what has been presented here. The study has several limitations, however. First, the study is dependent upon the quality and extent of data captured by each database that could underestimate the prevalence of some characteristics. For conditions and treatments, the assumption was made that patients did not have the health condition or treatment if it were not captured in the database. A specific example of this is prior systemic corticosteroid use that ranged from 7.5% to 28.4% in the COVID-19 hospitalized cohort and 5.1% to 26.4% in the diagnosed cohort. The prevalence of systemic corticosteroid use was slightly higher than the prevalence of an exacerbation

Figure 2. Prevalence of treatments among COPD patients with COVID-19 who have been diagnosed (red) and hospitalized (blue). *Databases contributing patients to both the diagnosed and hospitalized cohorts. SAMA=short-acting muscarinic antagonist.
SABA=short-acting beta2-agonist. LAMA=Long-acting muscarinic antagonist. LABA=Long-acting beta2-agonist. ICS=inhaled corticosteroids. NSAIDs=non-steroidal anti-inflammatory drugs.    Second, despite using a standardized data structure and method of analysis, heterogeneity between databases was still observed and it was not possible to determine whether this related to differences in clinical care compared to differences in the type of database. In this regard databases were a mixture between electronic health records and administrative claims capturing https://www.data.va.gov/ data from different health care settings. The definition of our hospitalized cohort meant that patients with hospital-acquired COVID-19 may have been included. These patients may have had a different disease course compared with communityacquired COVID-19 who were hospitalised due to developing severe COVID-19. We excluded patients with an asthma diagnosis prior to their index date. Despite this it is possible that some misclassification of asthma patients could still have occurred in older patients who were defined as having COPD using medications and age, if they were not previously coded as having asthma. In this regard, patients with asthma and COVID-19 may have a lower prevalence of comorbidities and differences in survival 26 . Our study analysed data specifically from the first wave of the COVID-19 pandemic. Since then, several COVID-19 variants have emerged with different severities and results may not be generalisable. Lastly, our study was descriptive in nature with the aim of being hypothesis generating and informing healthcare resource use. It was not designed to examine causal associations or to establish causal inference in relation to differences in COVID-19 outcomes that may result from differences in patient characteristics such as comorbidity that would require specialized study designs and regression analyses. However, this type of evidence has still contributed to support the understanding COVID-19 in patients in many settings and could be useful as hypothesis generating for future studies in patients with COPD 27 .

Conclusions
COVID-19 patients with COPD in the first wave of the pandemic were a vulnerable group with a high prevalence of other risk factors for severe COVID-19. No large differences in ICS use were seen between COPD patients with milder and more severe COVID-19 during the first wave of the pandemic although further studies are required to confirm or refute this. COPD patients experienced a high morbidity and mortality from COVID-19 during the first wave of the pandemic suggesting a multi-organ approach to clinical management was required.

Data availability
Underlying data Raw data from each database cannot be shared due to data privacy and governance requirements but raw data could be accessed according to the terms and conditions of each data source. The data source information including the terms and conditions for data access can be found in Table 8. Analyses were performed locally in compliance with all applicable data privacy laws. All aggregate data has been made freely available for public inquiry (https://data.ohdsi.org/Covid19Charac-terizationCharybdis/).