Questionaries where sent by postal office in October 1951 to 59,500 medical doctors. Each doctor was classified in one of 3 groups: current smokers, ex-smokers or non-smokers. Doctors that where current smokers and ex-smokers had to answer additional questions. Smoker doctors where asked about their smoking start age, quantity of tobacco they smoked and their method for consumption. Ex-smoker doctors had similar questions but related to their quitting of smoking. The lung cancer information was obtained from defunction certificates and other mortality data registered in subsequent years. The group of non-smoking doctors was constituted by those that informed that they had never constantly smoked as much as a cigarette per day for a year. The investigators received responses to their questionaries by 40,637 (68%) medical doctors, from which 34,445 where men and 6,192 where women.

The next section of this case study limits itself to the analysis of the male doctors with at least 35 years of age. The apparition of lung cancer in the doctors that responded the questionary was documented for a period of 10 years (November 1951 to October 1961) from the death certificated presented by the Secretary General of the UK and the lists of medical deaths given by the British Medical Association. All death certificated of the doctors where evaluated. For every death attributable to lung cancer, the medical registries where revised to confirm the diagnostic.

The lung cancer diagnostics where based in the best available evidence; around 70% had biopsies results, autopsies or cytology (combined with bronchoscopy or X-Ray evidence); 29% where cytology, bronchoscopy or X-Ray only; and only 1% where antecedents of cases, physical exams or death certificate. Out of the 4,597 deaths in the cohort in the period of 10 years, 157 where caused by lung cancer; 4 out of the 157 cases of that diagnostic couldn’t be documented, leaving 153 confirmed deaths by lung cancer.

Table 3: Number of deaths by lung cancer by daily cigarette consumption after the 1951 questionary (only for male doctors that weren’t smokers and current smokers). The number of smoked cigarettes was available for 136 individuals who died from lung cancer.

Question 1. Estimate the adequate association measure and confidence interval at 95% for every category of number of daily smoked cigarettes using the non-smokers group as reference.

Question 2. What can be inferred from the observed association in the different categories?

Question 3. What is the proportion of lung cancer deaths between all the smokers that can be attributed to tobacco? What is the proportion of lung cancer deaths in the whole population that can be attributed to tobacco? Consider that the lung cancer mortality in the total population in 0.94 x 1,000 years-person.
 

 

Number of cigarettes 0* 1-14 15-24 2 25 *Non-smokers Deaths 3 22 54 57 Person-years at risk 42,800 38,600 38,900 25,100