Nations with high smoking rate have low SARS-CoV-2 infection and low COVID-19 mortality rate.

Background and aim of the work The effect of tobacco smoking on COVID-19 disease is debated with common sense and experts suggesting a deleterious effect and manuscripts worldwide reporting a low prevalence of active tobacco smokers among intensive care unit patients. Methods We categorized countries worldwide into three groups with <25%; 25-45%; >45% of active male smokers with data expressed as median and interquartile range [IQR] and extracted data on SARS-CoV-2 infections and COVID-19 deaths per million inhabitants. We also applied multivariate regression techniques to adjust for several epidemiological factors. Results COVID-19 mortality was 13 (5-24) per million inhabitants in countries with male smokers >45% and 33 (4-133) in countries where male smokers were <25%. SARS-CoV-2 infection rates were 436 (217-954) and 1139 (302-4084) with data confirmed when dividing data for each continent and when controlling for confounding factors. Conclusions We found a counterintuitive low COVID-19 mortality and SARS-CoV-2 infection in countries with high prevalence of male smokers at the global level and within each continent, suggesting that active smoking habit is protective. Further research should urgently investigate which is the possible mechanism of action.

Introduction

As of October 6th 2020, >35 million SARS-CoV-2 cases were reported worldwide, with >1 million COVID-19 deaths[1]. Risk factors are still largely unknown, however, being male[2] and being active tobacco smoker[3] are surprisingly respectively associated with increased and reduced rates of SARS-CoV-2 infections and need for hospital and intensive care unit admission. We investigated whether countries with high prevalence of male smokers have a reduced rate of SARS-CoV-2 infections and COVID-19 deaths per million inhabitants.

Methods

SARS-CoV-2 infections and COVID-19 deaths per million inhabitants were extracted for all available countries on June 28th 2020[3] together with the percentage of adult male smokers in each country[4]. Countries with incomplete or missing data, either regarding the pandemic (n=14) or the smoking habits (n=91), were excluded. The remaining countries were categorized into three groups (with <25%; 25-45%; >45% of male smokers) with data expressed as median and interquartile range [IQR]. In order to explore the robustness of the unconditional results and to address potential omitted variable bias, we applied multivariate regression techniques (Ordinary Least Squares) to SARS-CoV-2 infections and COVID-19 deaths per million inhabitants as dependent variables. T-values on the estimated coefficients are robust to heteroskedasticity, i.e. to heterogeneity in the variance of the error term across countries. As explanatory variables we added (i) the starting date of the epidemics in each country; (ii) country population (expressed in millions), (iii) population density (number of inhabitants per square kilometer), (iv) the percentage of residents who were over 65 years old in the year 2017, to which we added in one specification real per capita Gross Domestic Product (GDP) in dollars in year 2017, adjusted for purchasing power parity to allow cross-country comparisons. The beginning of the epidemic in each country was defined as the day when the country reached 0.1 deaths per 1,000,000 population[5]. Total population and population density data were taken from the United Nations’ World Population Prospects 2019[6]. Data on the percentage of citizens >65 years and on real per capita GDP were taken from the World Development Indicators at the World Bank[7].

Results

The 110 countries included in the analyses are shown in table 1. COVID-19 median (IQR) mortality per million inhabitants was as low as 13 (5-24) in countries with male smokers >45% and as high as 33 (4-133) in countries where male smokers were <25% (figure 1). We found similar results for SARS-CoV-2 infection rates per million inhabitant: 436 (217-954) in countries with male smokers >45% and 1139 (302-4084) in countries where male smokers were <25% (figure 1). When splitting the data within each continent we found similar magnitude and direction of findings (Figures 2, 3 and 4).

Table 1.

list of all nations with available data on active male smokers, SARS-CoV-2 infection rate and COVID-19 deaths per million inhabitants (from World Health Organization(ref4), in decreasing order of male smokers’ rate):

Country	Male Smokers Rate %
Indonesia	76.2
Jordan	70.2
Russia	59
Georgia	57.7
Cuba	52.7
Greece	52.6
Armenia	52.3
Albania	51.2
Kyrgyzstan	50.4
Egypt	49.9
South Korea	49.8
Ukraine	49.4
Latvia	48.9
Bahrain	48.8
Montenegro	47.9
China	47.6
Bosnia and Herzegovina	47.2
Vietnam	47.1
Azerbaijan	46.5
Belarus	46.2
Moldova	45.7
Lebanon	45.4
Morocco	45.4
Mauritania	44
Kazakhstan	43.9
Serbia	43.6
Malaysia	43
Philippines	43
Bulgaria	42.4
Pakistan	41.9
Thailand	41.4
Estonia	41.2
Israel	41.2
Mauritius	40.1
Chile	40
Slovakia	39.7
Turkey	39.5
Croatia	39.4
Lithuania	38.1
Andorra	37.2
Nepal	37.1
Kuwait	37
Romania	36.9
Mali	36.8
Austria	35.5
Japan	33.7
Honduras	33.3
Germany	32.4
Poland	32.4
Hungary	32
Myanmar	31.6
Portugal	31.5
Mozambique	31.4
South Africa	31.4
Spain	31.3
Zimbabwe	31.2
Bangladesh	30.98
Bolivia	30.5
Jamaica	29.9
France	29.8
Malta	29.7
Argentina	29.5
Brunei	29.3
Czech Republic	29
Sri Lanka	28.4
Italy	28.3
Paraguay	28.3
Singapore	28
Saudi Arabia	27.9
Qatar	26.9
Switzerland	26.9
Uruguay	26.7
Belgium	26.5
Zambia	26.5
Netherlands	26.2
Luxembourg	25.8
Malawi	25.4
Uzbekistan	24.9
Kenya	24.6
Senegal	23.4
Finland	23.2
Ireland	22.4
Norway	22.4
Slovenia	22.3
Haiti	22.1
Iran	21.5
Peru	21.5
Oman	21
Mexico	20.8
India	20.4
Sweden	20.4
United Kingdom	19.9
United States	19.5
Brazil	19.3
Dominican Republic	18.8
Niger	18.6
Costa Rica	18.5
Benin	17.7
Canada	17.7
Denmark	17.6
Nigeria	17.4
New Zealand	17.2
Iceland	17
Australia	16.7
Colombia	16
Ecuador	14
Barbados	13.1
Ghana	13.1
Panama	10.6
Ethiopia	8.9

Figure 1.

SARS-CoV-2 infections and COVID-19 deaths per million inhabitants worldwide

Figure 2.

SARS-CoV-2 infections and COVID-19 deaths per million inhabitants in Europe (39 nations with available data), with different prevalence of active male smokers (<25%; 25-45%; >45%).

Figure 3.

SARS-CoV-2 infections and COVID-19 deaths per million inhabitants in the Americas (20 nations with available data), with different prevalence of active male smokers (<25%; 25-45%; >45%).

Figure 4.

SARS-CoV-2 infections and COVID-19 deaths per million inhabitants in Asia (32 nations with available data), with different prevalence of active male smokers (<25%; 25-45%; >45%).

In Table 2 we show that the negative correlation between mortality rates per million and prevalence of male smokers is strongly significant (at the 1% confidence level) and robust to controlling for confounding factors. More precisely, a percentage point increase in the prevalence of male smokers was associated with about 3 deaths per million less (columns 3 and 4). On the other hand, the negative correlation between total cases and prevalence of male smokers appeared to be less robust: it was mildly significant at the 10% confidence level (column 1) when not controlling for real per capita GDP, while it was estimated to be positive and non-significant at ordinary confidence levels when controlling for real per capita GDP (column 2).

Table 2.

Multivariate regressions on COVID-19 total cases and deaths.

Dependent Variable:	COVID-19 cases per million		COVID-19 deaths per million
	[column 1]	[column 2]	[column 3]	[column 4]
Prevalence of smoking males	-31.589*	18.875	-3.218***	-2.709***
	(-1.704)	(0.855)	(-3.836)	(-3.368)
Starting date of the epidemics	-53.210***	-2.698	-2.297***	-1.809**
	(-2.997)	(-0.186)	(-3.109)	(-2.493)
Population	-2.569**	-0.438	-0.033	-0.013
	(-2.191)	(-0.801)	(-0.786)	(-0.319)
Population density	0.680***	-0.371	-0.010**	-0.020***
	(3.031)	(-0.767)	(-2.485)	(-3.312)
Prevalence of over 65 citizens (2017 data)	-87.598	-233.191**	6.407***	5.044**
	(-0.985)	(-2.008)	(3.016)	(2.311)
Real per capita GDP (2017 data)	.	0.129**	.	0.001**
	.	(2.419)	.	(2.212)
Countries	106	105	106	105
R-squared	0.094	0.436	0.281	0.299

Notes: OLS (Ordinary Least Squares) estimates for the total number of confirmedCOVID-19 cases (columns [1] and [2]) and deaths (columns [3] and [4]). Heteroskedasticity-robust t-values are reported below each coefficient. *** denotes significance of the coefficient at 1% confidence level, ** for 5% confidence, and * for 10% confidence.

Regarding the other explanatory factors, it is interesting to note that the country-specific starting date of the epidemics was negatively and significantly correlated with mortality, thus confirming the previous result obtained by Landoni et al. on a smaller sample of 15 European and American countries[5]. Not surprisingly, the prevalence of citizens over 65 was positively and significantly correlated with COVID-19 mortality, while it was negatively and significantly correlated with total COVID-19 cases when controlling for real per capita GDP. Finally, real per capita GDP was positively and significantly correlated with both total confirmed cases and deaths.

Discussion

We found a low COVID-19 mortality and SARS-CoV-2 infection in countries with high prevalence of male smokers at the global level and within each continent, confirming[3] that active smoking habit is protective.

The reason why active smoking is protective against SARS-CoV-2 infection and its progression towards the most severe forms of the disease is unknown. Nicotine may play a major role by modulating the expression of the different isomers of the angiotensin converting enzyme receptors [ACE][8], including ACE-211, which is currently considered to be one of the entry doors for COVID-19 virus within the target cells[9].

Our work has several limitations, including its observational nature and the possibility that other non-included country-dependent factors might have influenced COVID-19 mortality.

In conclusion, a high male smoker prevalence was associated with low SARS-CoV-2 infections and COVID-19 deaths per million inhabitants worldwide and in each continent. Further research should urgently investigate which is the mechanism of action and whether surrogate techniques (with the obvious purpose of avoiding the drawbacks of tobacco smoking) could be used to prevent morbidity and mortality during the epidemic.

Copyright

The Corresponding Author had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The Corresponding Author declares that this manuscript is an original article and that he is able to prove its originality, if required from the Referees.

The Authors confirm that this manuscript has not been published before, either whole or in part, and that it is not currently under consideration for publication elsewhere.

The same Authors assign all rights, including but not limited to the copyright, for the present manuscript to Acta Biomedica upon its acceptance for publication.