Impact of smoking, COPD and comorbidities on the mortality of COVID-19 patients

The prognosis of the coronavirus disease 2019 (COVID-19) patients is variable and depends on several factors. Current data about the impact of chronic obstructive pulmonary disease (COPD) and smoking on the clinical course of COVID-19 are still controversial. This study evaluated the prevalence and the prognosis of COPD patients and smokers in a cohort of 521 patients admitted to four intermediate Respiratory Intensive Care Units (Puglia, Italy) with respiratory failure due to COVID-19 pneumonia. The prevalence of COPD and current smokers was 14% and 13%, respectively. COPD patients had a higher 30-day all-cause mortality than non-COPD patients. Former smokers compared to never smokers and current smokers had higher 30-day all-cause mortality. COPD patients and former smokers had more comorbidities. This study described the prevalence and the outcomes of COPD patients and smokers in a homogenous cohort of COVID-19 patients. The study showed that the prevalence of COPD and current smokers was not high, suggesting that they were not at increased risk of getting the infection. However, when SARS-CoV-2 infection occurred, COPD patients and former smokers were those with the highest all-cause mortality, which seemed to be mainly related to the presence of comorbidities and not to COPD and smoking itself.

www.nature.com/scientificreports/ There are several available studies evaluating for associations between smoking status and SARS-COV-2 infection 15 . A recent systematic review did not report smoking as a risk factor for SARS-CoV-2 infection but reported an increased risk of severe disease and need for mechanical ventilation or death in current smokers 16 . On the other hand, another meta-analysis did not find an association between current smoking and the severity of the disease 17 . Therefore, the impact of underlying COPD and smoking on patients' susceptibility and on the severity of SARS-CoV-2 infection is still uncertain and controversial 18,19 . In the present study, we hypothesized that COPD and tobacco smoking may have a different impact on patients' susceptibility to SARS-CoV-2 infection and on the severity and clinical course of the disease. The primary aim was to analyse the prevalence of COPD and the prognosis of COPD patients in a selected and homogenous cohort of patients with acute respiratory failure due to COVID-19-related pneumonia and admitted to intermediate Respiratory Intensive Care Units (RICUs), a model of care designed for monitoring and treating respiratory patients whose illness is at a level of severity that is intermediate between that which requires intensive care unit (ICU) facilities and that which can be managed on a conventional ward 20,21 . The secondary aim was to examine the prevalence of smokers and the association of tobacco smoking with sociodemographic and clinical features during the clinical course of these hospitalized patients.

Materials and methods
Study design. This is a retrospective, observational and multicentric study conducted in four intermediate RICUs of the Puglia region in Southern Italy (Policlinico Riuniti of Foggia, Policlinico of Bari, "F. Miulli" Hospital of Acquaviva delle Fonti and "A. Perrino" Hospital of Brindisi). We obtained medical records and compiled data from hospitalized adult inpatients affected by COVID-19 related pneumonia and acute respiratory failure, that were managed from March 5th to May 31st, 2020. A total of 521 consecutive patients with SARS-CoV-2 pneumonia, confirmed by RT-PCR tests on nasopharyngeal swab and by chest X-ray or chest CT performed in Emergency Rooms, were included in the study. At admission, all patients underwent to routine blood examination, assessing the following laboratory parameters: complete blood count, coagulation profile with D-dimers, serum biochemical tests (including renal and liver function, albumin, cardiac enzymes and iron), inflammatory markers [C-reactive protein (CRP) and erythrocyte sedimentation rate], lactate dehydrogenase (LDH), total lymphocytes and T-lymphocytes count, and vitamin D levels. The anamnesis, obtained both directly from the patients (if clinical conditions allowed) or from the relatives and/or clinical records, was aimed to ascertain smoking habits, the coexistence of respiratory diseases (such as asthma and COPD) and other comorbidities (diabetes mellitus, neurological diseases, kidney diseases, heart diseases, and cancer history). We used a semi-structured questionnaire in all the sites involved, investigating the same comorbidities. The anamnesis was also aimed at identifying patients on Long-term oxygen therapy (LTOT) or Long-term Non Invasive Ventilation (LTNIV) in order to assess the presence of chronic respiratory failure at baseline. Age-corrected Charlson Comorbidity Index (CCI) was calculated. The required ventilatory support and the arterial oxygen partial pressure to fractional inspired oxygen ratio (PaO 2 /FiO 2 ) ratio of each patient at admission were recorded. Since deaths occurred in hospital, the death data was obtained from the hospital databases. Data on 30 day-survival were also collected. On the basis of anamnestic data, the enrolled subjects were then divided into 5 groups to assess differences: (1) Current smokers with COPD, (2) Former smokers with COPD, (3) Current smokers without COPD, (4) Former smokers without COPD and (5) Never smokers. In the former smokers groups, we included those who had a history of smoking tobacco cigarettes in the past for at least 1 year and had stopped smoking for at least 3 months.
All patients' data were collected in the context of routine clinical care, and written informed consent was signed at admission according to hospital policy. Ethical Committee "Area 1 dell' Azienda Ospedaliero-Universitaria, Ospedale Riuniti" of Foggia approved this study. The experiment was performed in accordance with relevant guidelines and regulations. Statistical analysis. Continuous variables are presented as arithmetic mean ± standard deviation (SD), while categorical variables as counts (N) and percentage (%). Student T-test and ANOVA (with Turkey post-Hoc analysis) were used for comparing continuous data. Chi-square test (χ 2 ) was used to assess the relationship between categorical variables. The Kaplan-Meier method with hazard ratio (HR) was applied to detect differences in survival between the groups. A Cox proportional hazard model was used to estimate adjusted HRs with 95% confidence intervals (CIs) and to assess the influence of age, investigated comorbidities, CCI, and smoking habits. The cut off of the CCI was identified by a receiving operator curve (ROC) analysis, and a value greater than or equal to 4 was found to be the one with greater specificity (63%) and sensitivity (85%). A p value of < 0.05 was considered to be significant.

Results
A total of 521 COVID-19 patients were enrolled. Baseline characteristics of whole population are shown in Table 1. Data on smoking habits was available in 507 patients. According to a history of COPD and smoking, patients were divided into 5 groups (Table 2).
Out of 521 subjects, five required LTOT and none required LTNIV at baseline. Of the patients requiring LTOT, 2 had COPD (2.7%) and 3 were never-smokers without COPD.
Data on the ventilatory support required at admission were available only for 256 COVID-19 patients (Table 3). In this regard, the COPD group was significantly associated with the need to use mechanical ventilation in bilevel positive airway pressure (BPAP) mode [37 (16.97%) non-COPD patients vs. 15 (39.47%) of COPD patients underwent BPAP mode; p = 0.001].
The univariate Cox proportional hazard analysis of the 30-day relative risk of all-cause mortality on the whole population showed that age ≥ 65 years, smoking habits, CCI ≥ 4 and the presence of the different comorbidities investigated (except of history of cancer) increased all-cause mortality. However, after multivariate analysis, only age, smoking habits and some comorbidities (heart and neurological diseases) confirmed to be the greater risk factors of all-cause mortality, but not COPD (smoking habits: HR: 1.68, 95% CI: 1.34-2.11; age ≥ 65 years: HR:   (Table 4).

Discussion
To the best of our knowledge, this retrospective observational multicentre study was the first study that described the prevalence of COPD and the prognosis of COPD patients in a selected and homogeneous cohort of patients admitted to intermediate RICUs with COVID-19-related pneumonia. The present study also described the prevalence of smokers and the association of tobacco smoking with sociodemographic and clinical features during the clinical course of these hospitalized patients.
One of the most notable findings in our study was the higher prevalence of COPD (14%) in COVID-19 population compared with previous reports. In this regard, Attaway and colleagues showed that in a COVID-19 symptomatic population of 15.586 patients, there were no significant differences in the rate of SARS-COv-2 positivity between COPD and non-COPD patients 22 . However, the study itself showed a higher prevalence of COPD patients that required hospitalization (48.2% of COPD population vs. 26.5% of non-COPD population) 22 . Other studies reported a COPD prevalence of 2% in COVID-19 patients 23 . Emami and colleagues showed a COPD incidence of 0.95% in hospitalized COVID-19 patients 24 . The Spain registry of hospitalized patients with COVID-19 showed a relatively low COPD incidence of 7.7% 25 . Our previous study conducted on 97 hospitalized COVID-19 patients also showed a low percentage of respiratory comorbidities (8.3%) and therefore of COPD 12 . To date, only the review article of Leung and colleagues has suggested an extremely variable prevalence (from 1.2 to 38%) of COVID-19 patients with COPD, which is the closest to the present finding 26 . A possible explanation of our finding concerns the higher age of our population (mean age 67.18 ± 17.5 years). One another possible explaining hypothesis is that reduced use of drugs for respiratory diseases, such as inhaled corticosteroids or bronchodilators, might have favoured the infection. In fact, corticosteroids could represent a protective factor   Whether COPD could represent a risk factor for negative prognosis in COVID-19 remains to be clarified. However, when we talk about evidence in this area, we must consider that the race to publish, especially in the first phase of the pandemic, may have led to publication of studies with a lot of methodological biases (e.g., poor statistical analysis, non-standardized data collection methods, non-homogeneous populations, too small simple sizes, unadjusted prevalence data) and to erroneous conclusions, which may be responsible for current controversies. In our study, COPD was associated to a major requirement of non-invasive ventilation in BPAP mode (16.97% of non-COPD patients vs. 39.47% of COPD patients underwent BPAP mode; p = 0.001) and to a higher 30-day all-cause mortality (non-COPD patients: 21% vs. COPD patients: 52%; p = 0.0001). These findings are in agreement with those of Attaway and colleagues showing that higher rates of ICU admission [adjusted odds ratio (OR): 1.20) and invasive mechanical ventilation (adjusted OR: 1.49) were observed in COPD patients compared with non-COPD patients in a COVID-19 population 22 . Guan and colleagues described similar findings, as they showed that COPD patients were more likely to experience ICU admission, invasive ventilation, and death 8 . Taken together, these data confirmed that COPD could be a risk factor for severe COVID-19. But, how can COPD be responsible for more severe forms of COVID-19? An intriguing possibility concerns the ACE-2 enzyme, which is essential for the virus to gain entry into the organism and it is overexpressed in the lower respiratory tract of COPD patients, favouring the severity of the disease 27 . Another hypothetical explanation is the alteration of the renin-angiotensin-aldosterone system in COPD patients, which could lead to pulmonary edema during SARS-CoV-2 infection 28 . The presence of patients with severe forms of COPD at baseline might be a further explanation. In particular, in patients with severe COPD without COVID-19 the presence of chronic respiratory failure and, therefore, the use of LTOT, is associated with negative consequences which might also be responsible for the poor prognosis of patients with severe COPD and COVID-19 29,30 . However in our population only 2 (2.7%) COPD patients with COVID-19 had a history of LTOT/chronic respiratory failure, not explaining the worst prognosis of these patients. By contrast, in the present study, the higher age and the presence of comorbidities in COPD patients may explain the greater severity of their disease. In fact, while the univariate analysis on our sample showed that age, smoking, individual comorbidities and CCI ≥ 4 increase the 30-day relative risk of allcause mortality, the multivariate analysis revealed that it was not COPD itself the cause of the greater severity of COVID 19, but are the higher age, the smoking habits, and the presence of other comorbidities (in particular cardiovascular and neurological ones). Since CCI is an index that includes both age and comorbidities, it could represent a valid tool for predicting the prognosis of these patients. These findings are in agreement with those of Gómez Antúnez and colleagues obtained from a study conducted on 10,420 patients 13 . In particular, the association between multimorbidity and negative prognosis of COPD patients had already been widely emphasized by Almagro and colleagues in the field of COPD exacerbation. Their most recent studies suggested that 3-month mortality in hospitalized COPD patients was correlated to a higher CCI 31,32 . Therefore, the increased presence of multimorbidity may be responsible for the worse prognosis in COVID-19 patients affected by COPD.
Another important point of discussion of the present study was to understand whether current smokers are more prone to contract SARS-CoV-2 infection. Since several studies in the past have shown that smokers are predisposed to severe respiratory infections, it would be logical to expect a high prevalence of current smokers in hospitalized COVID-19 patients 33 . Oppositely, in our population of patients hospitalized with COVID-19 pneumonia, we found a prevalence of current smokers of 13%, which is low if compared with the 20.3% prevalence of smokers in the 60-64 age class and with the 14.4% prevalence in the 65-74 age class of the general Italian www.nature.com/scientificreports/ population 34 . This surprising and paradoxical finding is in line with most similar studies. In particular, a Chinese analysis conducted on hospitalized COVID-19 patients showed a low prevalence of smokers (6.5%) compared to that of the general Chinese population 35 . Miyara and colleagues revealed a low incidence of current smokers even in a COVID-19 French population 36 . On the contrary, Guan and colleagues found a higher prevalence of current smokers in severe forms of the disease than in milder ones, suggesting that the prevalence may vary depending on the severity of the examined population 6 . In this regard, we would like to point out that, although our population was mostly made up of severe patients with acute respiratory failure, the prevalence of current smokers remained low. Also Cumming and colleagues reported a low prevalence of smokers in critically ill patients with COVID-19 37 . The low rate of smokers observed in COVID-19 patients suggests that smokers are less prone to contract SARS-CoV-2 infection and that smoking habit has a "protective" effect. Several mechanisms may explain how this "smoker's paradox" may be possible 38,39 . A possible mechanism concerns the nitric oxide produced during smoking, which inhibits the replication of the virus and its entry into cells. Another explanation applies the reduced expression of ACE-2 in smokers, which could reduce the risk of attachment of the virus to upper respiratory tract cells and the consequent infection 19 . However, the reduced expression of ACE-2 itself may be harmful, as it exposes the smoker to a greater risk of local inflammation, vasoconstriction and thrombosis worsening his outcome 19 .
In fact, about the link between smoking and clinical outcomes of COVID-19, a systematic review concluded that smoking was associated with disease progression and increased adverse outcomes in COVID-19 16 . Similar findings derived from a very large study of the UK general population, where smoking was significantly associated with increased COVID-19 mortality after age and sex adjustment (OR: 1.25, 95% CI: 1.12-1.40). However, after adjustment for multiple additional covariates, the same study found that smoking was associated with a reduced risk for COVID-19 mortality (OR: 0.88, 95% CI: 0.79-0.99) 40 . Moreover, other studies in particular analysed the relative risk of current smokers. Lippi and Henry's meta-analysis demonstrated no significant association between current smoking and COVID-19 severity (OR: 1.69, 95% CI: 0.41-6.92; p = 0.254) 17 . On the other hand, some reviews have highlighted an increased mortality and need for mechanical ventilation in current smokers 15 . Due to these controversies, different mechanisms by which smoking could exert a protective or harmful action on the clinical course of COVID-19 have been proposed. A possible protective mechanism of smoking relates to the inhibition of pro-inflammatory cytokines production by nicotine, protecting against cytokine-storm syndrome. Another theory concerns the continuous suppression of these systemic cytokines in smokers, who may adapt their immune response by becoming more tolerant and less reactive to continuous inflammatory stimuli than patients who have never smoked. However, smokers have a higher prevalence of comorbidities, that are known to have an independent negative impact on the clinical course of COVID-19 19 . A very important and innovative finding of the present study was that, among smokers, former smokers were those with higher mortality. Since former smokers had higher multimorbidity than never smokers and current smokers, and since the multivariate analysis showed comorbidities like neurological and cardiovascular ones among the risk factors for mortality, we can hypothesize that it was not smoking itself the responsible for the worst prognosis of smokers, but the comorbidities related to its chronic exposure.
The present study had some limitations. Firstly, since patients were often critically ill at admission, pulmonary function tests could not be performed. Therefore, the diagnosis of COPD was based on purely anamnestic data. Secondly, we analysed data basing on what was available in the hospital databases, but we recognized there were some information not captured by them, including knowledge about existing medications, through which we would have confirmed or not the diagnosis of COPD. Furthermore, we did not have cause-specific data on mortality. Moreover, given that the present study analyzed outcomes from March 5th to May 31st 2020, therefore before corticosteroids were routinely used, the lack of this medication in the first phase of the pandemic may have influenced all-cause mortality in our hospitalized COVID-19 patients. Finally, we did not have a cumulative effect of smoking over time, as measured by pack-years. In fact some studies have suggested that the cumulative exposure to cigarette smoking was an independent risk factor for hospital admission and death from COVID-19 41 .

Conclusions
Based on our study findings, in a selected and homogeneous cohort of hospitalized COVID-19 patients, the prevalence of COPD was not high, suggesting that they were not at increased risk of getting the infection. However, when SARS-CoV-2 infection occurred in COPD patients, they had a higher all-cause mortality and a worse prognosis than non-COPD patients. The present study also showed that SARS-CoV-2 infection was not common in current smokers and that, among smokers with COVID-19, former are those with the higher mortality. The worst outcomes of COPD patients and former smokers seemed to be mainly related to the comorbidities of these subjects and not to COPD or smoking itself. www.nature.com/scientificreports/