Low rate of daily active tobacco smoking in patients with symptomatic COVID-19

Importance: As the pandemic of COVID-19 is still under progression, identification of prognostic factors remains a global challenge. T he role of smoking has been suggested among the disease risk factors, although it is highly controversial. Objective: T o evaluate whether the rate of daily smokers in patients with COVID-19 was different to that in the French population. Participants: COVID-19-infected inand outpatients in a large French university hospital between February 28, 2020 and March 30, 2020 for outpatients and from March 23, till April 9, 2020 for inpatients. Design: We systematically interviewed the patients on their smoking status, use of ecigarette and nicotinic substitutes. T he rate of daily smokers in inpatients and outpatients were compared to those in the 2018 French general population, after standardization for sex and age.


Introduction
As the pandemic of COVID-19, caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is still under progression, the identification of risk factors is a global challenge. Among epidemiological risk factors, the role of smoking, to date, is unclear. Smoking has been initially found associated with adverse disease prognosis of COVID-19 [1], although this finding remains controversial [2]. Reported rates of current smokers among SARS-CoV-2-infected patients range from 1.4% to 12.5% in China [1,[3][4][5][6][7][8][9][10], from 1.3% to 5.1% in the USA [11,12], mainly for hospitalized patients (see systematic review in [13]). Very recently, results of an observational database from 169 hospitals in Asia, Europe, and North America involving 8,910 COVID-19 hospitalized patients reported rates of current smokers of 5.6% in North America, 5.4% in Europe and 5.9% in Asia [14]. For outpatients, data are very scarce but also suggest similar low rates [13]. At first approach, the rates of current smokers in both COVID-19 in-and outpatients seem to be low compared to the general population. T hese data notwithstanding, no firm conclusions can be drawn from these available COVID-19 studies because main potential confounders for smoking rate, namely age and sex, were not taken into account. Additionally, these studies included mostly hospitalized patients, and the low rate of current smokers may be related to high rate of patients with comorbidities (smokers having been advised to quit). Furthermore, these studies used data collected in the context of care in the medical files, which favors underreporting (patients being considered as non smokers when smoking status is not reported in the medical file) particularly when data collection is made by overwhelmed care healthcare teams for a disease a priori not related to smoking, and biased reporting (preferential smoking status collection in patients with pulmonary or cardiovascular comorbidities).
T herefore, the effect of current smoking on the risk of SARS-CoV-2 infection has yet to be determined. T o accurately evaluate whether or not current smoking is associated with the risk of COVID-19, we conducted an observational study specifically designed to T his study is observational and has been approved by the ethics committee of Sorbonne University (N° 2020 -CER-2020-13).

Definitions and data collected
Confirmed COVID-19 was defined as a positive result on real-time reverse-transcriptasepolymerase-chain-reaction (RT -PCR) assay of nasal and pharyngeal swab specimens.
Smoking status was collected in all patients by specifically asking whether they were current smokers (and if so, to provide details on their smoking habits: daily or occasional smoking, number of daily cigarettes), former smokers, or not smokers ever. We used the same definition as in the French national annual survey of smoking habits (Santé Publique France Health Barometer) [15]. Daily smokers were defined as individuals reporting daily smoking of cigarettes (manufactured or rolled) or other tobacco products (cigars, cigarillos, pipe, shisha). Occasional smokers were defined as individuals reporting infrequent, but not daily smoking. T he group of former smokers included anyone having smoked in the past, occasionally or daily, and had abstained from smoking prior to COVID-19 onset. T he term "never smoker" designated people who had never smoked.
In addition, for all outpatients and for all inpatients, we systematically asked former

Statistical analysis
A descriptive analysis has been made by group (inpatients -outpatients). Qualitative variables were described by numbers and percentages, and quantitative variables by median and interquartile range. Inpatients and outpatients were compared for age and sex with Wilcoxon test and Pearson Chi2 tests, and for comorbidities and smoking status by logistic regression adjusted on age and sex. T he SIRs were used to compare daily smoker rates in the COVID-19 inpatients and outpatients, respectively, with those of daily smokers in a reference population, here the French general population in 2018. T he estimated SIR and its 95% confidence interval is the ratio between the observed number of daily smokers among the COVID-19 patients and the number of daily smokers that would be expected in the study population, on the basis of age-and gender-specific current daily smokers rates in the general population. T he main analysis involved all included patients, and those older than 75 years were considered in the 65-75 years age class for standardization, which for our hypothesis is a conservative approach, because daily smoker rates decreases with age. For 7 outpatients and 2 inpatients, we were unable to interview the patient on his smoking status. We did not include the latter patients in the main analysis because the missing smoking status was very likely to be at random (7 outpatients that could not be reached, and among the 2 inpatients, one due to the language barrier and the other due to severe cognitive impairment). We performed two sensitivity analyses, one excluding patients older than 75 years, the other considering the patients with missing smoking status as daily smokers.
We also estimated the SIR in healthcare workers and non healthcare workers in the outpatients (as healthcare workers were overrepresented, because they were tested at their workplace in case of symptoms).

Demographic and Clinical Characteristics
A total of 340 inpatients and 139 outpatients were included. T he demographic and clinical characteristics of the two groups are shown in T ABLE 1. As shown in FIGURE 1, age distribution differed between outpatients and inpatients, with outpatients being younger and inpatients older.  Occasional smokers were 6 (4.5%), 2 have stopped smoking since COVID-19 onset and none has taken nicotinic substitutes. Former smokers were 41 (31,1%; 21 men and 20 women). Among these, 2 (4.9%) had quitted three months before COVID-19 symptoms onset and 39 (95.1%) more than 1 year before; 2 (4.9%) were using nicotinic substitutes (1 by use of e-cigarette). Among the 77 non-smokers, none were using nicotinic substitute (data was missing for 7).
T he comorbidities were more frequently observed in inpatients than in outpatients:   T he SIRs did not differ between outpatients and inpatients, suggesting that the potential effect of smoking is towards symptomatic COVID-19, irrespective of the severity. In the rare daily smokers in the COVID-19 patients of our study, we did not observe any effect of the daily cigarette consumption. Actually some were heavy smokers and others not.
T o note, in 2018, the mean number of daily cigarettes by current smokers in the French general population was 13.0 cigarettes, or equivalent, with 14.0 cigarettes for men and 11.9 for women [15]. We also observed a very rare use of nicotinic substitutes in the former smokers (2/111 in the inpatients and 2/41 in the outpatients, one of each group with e-cigarette), and in none of the outpatients non smokers, which is in line with the national survey indicating that e-cigarette use is still low in France (3.8% of daily users), and does not concern non smokers (1% of e-cigarette users) [15]. compared to a control group or the general population except in two, where the current smokers' rate in the general population is reported, with no statistical comparison and thus not accounting for the age and sex distribution of the COVID-19 patients.
Our findings are in line with those from Fontanet et al. 2020 [18], who reported smoking habits in a cohort of pupils, their parents and siblings, as well as teachers and nonteaching staff of a high-school located in Oise (n = 661). Smokers had a lower risk of confirmed COVID-19 (as defined by antibodies detection) compared to non-smokers (7.2% vs 28.0; age-adjusted OR = 0.23; 95% CI = 0.09 -0.59), and the association was also significant after adjustment on occupation.
Our study has many strengths. By contrast, with previously reported studies, our study was specifically designed to assess smoking habits in the Covid-19 patients. Previous studies used smoking status as recorded in the medical files, which are subject to underreporting (usually not accounted as missing data) and biased reporting. In our study, patients were systematically interviewed about their smoking habits, and use of nicotinic substitutes. T he rate of missing data -one of the more frequent caveat of studies reported so far -was very low (1.9%). Additionally, to completely rule out the impact of missing data on the conclusion of our study, we did a sensitivity analysis, considering that patients with missing smoking status as daily smokers, which is conservative regarding the hypothesis of a protective effect of smoking. In this sensitivity analysis, the SIR remained significantly below 1 showing the robustness of our results. Furthermore, we used the same definitions as the French national annual survey of smoking categories (Santé Publique France Health Barometer) [15] that we used for reference to calculate the SIR. Finally, we investigated apart the association of daily smoking with COVID-19 separately in outpatients and inpatients, which provides relevant information in addition to previous studies.
Our study has also several limitations. First, the study was performed in early 2020 and the reference smoking rates in France were estimated from January to June 2018, as Second, because patients hospitalized in ICU were not included in the present study, we could not conclude whether daily current smoking was associated with very severe forms of COVID-19. Furthermore, as the rate of daily smokers was very low in both out-and inpatients, the study was not powered enough to assess whether smoking was associated with severity as defined by being hospitalized. However, it provides the information of a low smoking rate of daily smokers even in COVID-19 outpatients, which is of great interest in the understanding of the phenomenon, because it shows that smoking appears to have an effect at least on the risk of infection. Previous studies did not make it possible to decide between two hypotheses: a protective effect on the decrease of the risk of infection or on the decrease of the risk of severe form of infection (as defined by hospitalization for COVID -19). T he association between daily smoking and COVID-19 severity still remains controversial [14]. A larger well-designed study including also ICU patients will certainly help to conclusively address this question.
However, collecting accurate smoking status is difficult in ICU patients.
T hird, smoking status was self-reported by the patients, which tend to underestimate daily smokers' rate due to social desirability bias [20]. However, we used the same methodology as the Baromètre Santé survey that we used as reference. Furthermore, in the French healthcare system, access to care is not rationed based on any potential for positive outcome, or compliance with Public Health recommendations, thus there may be no particular incentive to underreport being a current smoker.
Finally, in our study, smoking status was assessed only in symptomatic COVID-19 patients while a part of infected individuals are asymptomatic [21]. T hus, we cannot conclude whether daily smoking is associated with SARS-CoV2 infection, or to symptomatic forms of this infection. T he recent study by Fontanet [18], which highlights a decrease in the risk of COVID-19 of the same order of magnitude as us, gives a key to answer to this question because this study, based on serological results, takes into account both symptomatic and asymptomatic forms.
Because this is a cross-sectional study, we cannot confirm the causality of the association. We cannot also identify which of the many compounds of tobacco exerts the protective effect of smoking on COVID-19. T here are however, sufficient scientific data to suggest that smoking protection is likely to be mediated by nicotine. SARS-CoV2 is known to use the angiotensin converting enzyme 2 (ACE2) receptor for cell entry [22][23][24], and there is evidence that nicotine modulates ACE2 expression [25] which could in turn modulate the nicotinic acetyl choline receptor [26]. We hypothesize that SARS-CoV2 might alter the control of the nicotine receptor by acetylcholine. T his hypothesis may also explain why previous studies have found an association between smoking and COVID-19 severity. [1,3,6] As hospitals generally impose smoking cessation and nicotine withdrawal at the time of hospitalization, tobacco (nicotine) cessation could lead to the release of nicotine receptors, that are increased in smokers, and to a "rebound effect" responsible for the worsening of disease observed in hospitalized smokers. However, this hypothesis needs further investigation, and the deleterious role of smoking in hospitalized patients with COVID-19 cannot be ruled out to date.
In conclusion, our results suggest that active smokers may be protected against symptomatic COVID-19. T his was evidenced for outpatients (who have less serious infections) as well as for hospitalized patients. T he physiopathological process underlying this effect may involve nicotine through the nicotinic receptor (and not the smoke of cigarettes per se), an hypothesis which deserves further evidence. In light of the possible increased risk of severe form of COVID-19 among smokers once infected and of the long-term harmful consequences of smoking which is responsible for a very heavy public health burden with more than 78,000 deaths per year in France, our findings needs careful consideration and cannot be translating into a clinical practice as it. Careful investigation of the potential protective effect of nicotine should be investigated both in in vitro and in vivo before any firm conclusion can be drawn.