Data That I Correlate: Is smoking a factor in COVID-19 deaths?

I know I harp on about smoking.  But when we’re talking about a pandemic that is literally choking people to death and damaging lungs, how can you ignore the possible connection?

Worldometers has become a commonly cited source on COVID-19 for number of infections, deaths and recoveries, providing daily updates. The site is run by a company a Chinese company Dadax, but the site’s “About” page shows where it’s referenced and recommended by other sites (e.g. American Library Association). The numbers of cases and deaths in the table below come from the Worldometers site (copied at time of writing).  The fatality rate is not given, but is easily calculated.  The percentage of smokers by country comes from World Population Review.  (Iraq’s rate of men who smoke is approximately 30%.)

Coronavirus Cases: 338,225    
Deaths: 14,457    
Recovered: 96,958    
Country,       Total     Total     Deaths    % of    
Other          Cases     Deaths    by %      Smokers
----------     ------    ------    ------    -------
China          81,054     3,261    4.02%     24.70%
Italy          59,138     5,476    9.26%     24.00%
USA            38,167       396    1.04%     17.25%
Spain          28,603     1,756    6.14%     29.20%
Germany        24,806        93    0.37%     30.35%
Iran           21,638     1,685    7.79%     11.10%
France         14,459       562    3.89%     27.70%
S. Korea        8,897       104    1.17%     27.00%
Switzerland     7,474        98    1.31%     23.30%
UK              5,683       281    4.94%     19.15%
Netherlands     4,204       179    4.26%     25.05%
Belgium         3,401        75    2.20%     23.25%
Austria         3,302        16    0.48%     35.15%
Norway          2,263         7    0.31%     22.25%
Sweden          1,931        21    1.09%     20.60%
Portugal        1,600        14    0.88%     22.60%
Canada          1,426        20    1.40%     14.95%
Denmark         1,395        13    0.93%     17.00%
Australia       1,353         7    0.52%     14.90%
Malaysia        1,306        10    0.77%     22.20%
Brazil          1,209        18    1.49%     15.30%
Czechia         1,120        NA              NA
Japan           1,086        36    3.31%     22.15%
Turkey            947        21    2.21%     25.95%
Israel            945         1    0.11%     30.25%
Ireland           906         4    0.44%     22.15%
Luxembourg        798         8    1.00%     23.60%
Ecuador           789        14    1.77%      8.65%
Pakistan          646         4    0.62%     22.45%
Chile             632         1    0.16%     38.00%
Poland            627         7    1.12%     28.05%
Finland           626         1    0.16%     20.85%
Greece            624        15    2.40%     42.65%
Thailand          599         1    0.17%     21.85%
Iceland           568         1    0.18%     16.05%
Indonesia         514        48    9.34%     39.90%
Saudi Arabia      511        NA              15.40%
Qatar             481        NA              NA
Singapore         455         2    0.44%     16.50%
Romania           433         2    0.46%     29.80%
Slovenia          414         2    0.48%     20.20%
India             396         7    1.77%     11.15%
Philippines       380        25    6.58%     25.75%
Russia            367         1    0.27%     40.90%
Peru              363         5              NA
Bahrain           332         2    0.60%     28.20%
Estonia           326        NA              33.05%
Hong Kong         317         4    1.26%     NA
Egypt             294        10    3.40%     25.10%
Croatia           254         1    0.39%     36.45%
Mexico            251         2    0.80%     13.70%
Lebanon           248         4    1.61%     38.20%
Panama            245         3    1.22%      6.60%
South Africa      240        NA              18.95%
Iraq              233        20    8.58%     NA

There is a noticeable trend. In most countries with high rates of smoking and at least 10 deaths, the mortality rate from COVID-19 is over 2%. (The world national average of smokers is 20%.)  The only countries with a mortality rate over 2% and a smoking rate below the world average of smokers are the UK (19.15%) and Iran (11.10%). They are not outliers since UK smoking rates were over 20% until 2010 and Iran and Italy are both part of China’s “belt and road initiative”, many people commuting between the countries.

China, Italy, Spain, France, UK, Netherlands, Belgium, Greece, Japan, Turkey, Indonesia, Philippines and Egypt all have smoking rates over 20% and over 2% of their infected have died.  Few countries buck that trend (many smokers but less than 2% fatality rate: South Korea, Switzerland, Austria, Germany, Malaysia, Portugal).  Most have advanced medical care systems or were prepared when the first cases appeared.

Worldometers groups all people into two genders, and gives no information on Transgender and Non-Binary people.  There is a much higher fatality rate among “males” than “females” from COVID-19. The WHO claims 40% of men smoke worldwide, only 9% of women, so if smoking is involved, it follows the pattern.

Death Rate by sex:

          Confirmed      All
Sex       cases          cases
------    ---------      -----
Male      4.7%           2.8%
Female    2.8%           1.7%

If smoking is a factor in the mortality rate, then Indonesia is the country at greatest risk. Smoking rates among Indonesian men is 76%, and before COVID-19 appeared 600 smokers die daily in Indonesia from cancer, lung disease and heart disease (youtube video by Al Jazeera). The country has had an estimated 500,000 COVID-19 exposures. Most income earners are men and there is no government social welfare system.

I know, I know, correlation does not prove causation, and I’m not a medical professional. But COVID-19 is a disease that attacks and damages the lungs. I find it hard to believe that smokers and non-smokers will be infected and die at the same rate, even among younger smokers.


  1. kestrel says

    This is interesting; the Partner thinks what matters most in this is comorbidities. Maybe smoking could be considered that way? I’m not sure. Mostly the Partner is talking about things like diabetes or COPD, for example, and has noted that those with comorbidities tend to die with this disease. We have a friend with COPD and are terrified he’ll contract coronavirus; we don’t think he would pull through.

  2. lochaber says

    I’m thinking something similar to what kestrel said. I’m not certain that smoking would make one more likely to be infected by COVID19, but that someone with already existing lung damage from smoking and such may be more likely to suffer more severe effects, and have a higher risk/rate of death than someone with relatively undamaged lungs.

    not a medical professional, and all that…

  3. kestrel says

    OK just talked to the Partner (who works in the ER at a hospital, and who’s claim to fame is being able to get an IV on very difficult patients) and this is what I learned. They apparently triage people according to whether or not they smoke – if they do smoke, it’s something they have to take in consideration when treating the patient. Also apparently smoking a cigarette caused the veins to constrict, making it far more difficult to get an IV on that person. The Partner can apparently tell if the patient smoked a cigarette before coming in to get an IV started… and has startled patients with this apparently “psychic” ability.

    So you might be on to something here. Good research!

  4. says

    I certainly wasn’t saying smoking would cause COVID-19. But why wouldn’t it be considered an increased risk and grouped with those who have diabetes, cancer or other pre-existing conditions?

  5. Owlmirror says

    The high smoking rate and low mortality from COVID in Russia, Lebanon, Romania, Chile, and Croatia, makes me wonder if there’s something wonky in the diagnoses/reporting, like maybe they’re attributing deaths to other pulmonary factors or something. Hm.

  6. brikoleur says

    Please don’t publish this type of armchair analysis unless you have an understanding of what the numbers mean.

    The CFR numbers you’re seeing at this point mostly reflect one thing: how comprehensively people are tested, and who is being tested. If you’re only testing patients with critical double pneumonia and a negative on an influenza test, you will get a very high CFR. If you’re testing everyone who so much as sneezes like South Korea style, you will get a very low CFR. In other words, your correlation is between test policy and smoking, not infection/fatality rate (IFR) and smoking. There might be one of course but it’s not in your data.

    Here is a very recent analysis that attempts to estimate the success various countries have had in catching COVID-19 cases, based on the ratio between confirmed cases and deaths:

    Here is a Twitter list of actual researches you might want to follow:

    • jrkrideau says

      That first link looks interesting though not particularly relevant to Intransitive’s post and the second seems essentially gibberish seen in relation to the post.

      Keep up the good work.

      • brikoleur says

        The Net is full of amateur epidemiologists making bad inferences from public data. This one is harmless – I’d be amazed if smoking isn’t a risk for COVID-19, so yeah now would be a terrific time to stop if you’re a smoker – but a lot of it isn’t, like that one “everything is fine” Medium post that got widely quoted before it was yanked. But I do feel quite strongly that it does not help to contribute to the mass of these bad inferences, it reduces the signal/noise ratio and does not enlighten anyone.

        The data is extremely noisy to start with. To make any meaningful comparisons between them, you need to understand what numbers are comparable in the first place, and for what purpose. If you don’t even know the difference between CFR and IFR, how any of these numbers are estimated or inferred, what the uncertainties related to them are, what factors go into them, etc., you’re gonna find a lot of entirely specious correlations. For example, if you actively look for them and compare the right numbers, you’ll find datasets where smokers are underrepresented. Somebody tried to convince me that smoking protects against corona just this way the other day.

        If you want to dig into how this stuff works, there are worse places to start than this:

  7. StevoR says

    Not a doctor but couple of things hopefullyworth noting here :

    Scrolldown to #6

    In Indonesia some think smoking can prevent someone getting it
    As told by Indonesia correspondent Anne Barker

    Who is saying this? Social media users in Indonesia.

    What are they saying? Posts on social media in the past few weeks have claimed coronavirus does not attack people who smoke because the composition of tobacco and cloves can resist the attack.

    Which is wrong, of course, and the experts here debunk it quickly :

    Professor Amin Soebandrio.

    He says smoking increases ACE 2 receptors in the lungs that cause the COVID-19 virus.

    Professor Soebandrio says each receptor acts like a port, so if there are more berths, more “ships” will come.

    Professor Oliver agrees, saying if anything “smoking makes the outcomes worse”.

    Also apparently Indonesia has the highest detah rate from COVID 19 onthe planet – see :

    with it being likey that theer are alot mroe cases of infection there than we currently know about.

    Incuidentally it seems India is one nation that is doing better than expected here :

    Indeed quite plausibly better than here in Oz or inthe USA.

  8. StevoR says

    Argh! Apologies for all the typos – reallywa shalf asleep last night writing that.

    Clarifying fixes :


    hopefully worth / scroll down / on the planet

    .. with it being likely that there are a lot more cases of infection there than we currently know about./ Incidentally.

    Guess you could tell from context. My apologies again.

  9. says

    Mon français est terrible.

    I was expecting naysaying. I’ve seen another article saying the male-female disparity is genetic or hormonal. Female mice with hysterectomies died more often from viruses while male mice give estrogen died less. It’s not proof, but it follows the observation. This is not the item I first saw, but it will do.

    Now I’m in a quandry: Do I start taking estrogen again and risk a heart attack or stroke? Or not take it and risk death if COVID-19 gets loose in Taiwan?

  10. Owlmirror says

    This is mostly about Spain, but it also has this paragraph:

    In Italy, authorities have conceded that their coronavirus death toll did not include those who had died at home or in nursing homes. Similarly in France, officials have said that only those who died in hospitals had been recorded as pandemic-related — a practice they said would change in the coming days.

    So even in Italy and France, the reporting is problematic.