1. ¹MRC/CANSA/NHLS/WITS Cancer Epidemiology Research Group, National Cancer Registry, PO Box 1038, Johannesburg 2000, South Africa
2. ²Department of Anatomical Pathology, National Health Laboratory Services (formerly South African Institute for Medical Research) and University of the Witwatersrand, York Road, Parktown 2196, South Africa
3. ³Department of Radiation Therapy, University of the Witwatersrand, York Road, Parktown 2196, South Africa
4. ⁴Department of Medical Oncology, University of the Witwatersrand, York Road, Parktown 2196, South Africa
5. ⁵Department of Haematology, University of the Witwatersrand, York Road, Parktown 2196, South Africa
6. ⁶Cancer Research UK, Epidemiology Unit, Oxford, UK

Correspondence: F Sitas, E-mail: freddys@mail.saimr.wits.ac.za

Received 21 September 2001; Revised 13 March 2002; Accepted 3 April 2002.

Abstract

The authors used data collected from 1995 to 1999, from an on-going cancer case–control study in greater Johannesburg, to estimate the importance of tobacco and alcohol consumption and other suspected risk factors with respect to cancer of the oesophagus (267 men and 138 women), lung (105 men and 41 women), oral cavity (87 men and 37 women), and larynx (51 men). Cancers not associated with tobacco or alcohol consumption were used as controls (804 men and 1370 women). Tobacco smoking was found to be the major risk factor for all of these cancers with odds ratios ranging from 2.6 (95% CI 1.5–4.5) for oesophageal cancer in female ex-smokers to 50.9 (95% CI 12.6–204.6) for lung cancer in women, and 23.9 (95% CI 9.5–60.3) for lung cancer and 23.6 (95% CI 4.6–121.2) for laryngeal cancer in men who smoked 15 or more grams of tobacco a day. This is the first time an association between smoking and oral and laryngeal cancers has been shown in sub-Saharan Africa. Long-term residence in the Transkei region in the southeast of the country continues to be a risk factor for oesophageal cancer, especially in women (odds ratio=14.7, 95% CI 4.7–46.0), possibly due to nutritional factors. There was a slight increase in lung cancer (odds ratio=2.9, 95% CI 1.1–7.5) in men working in 'potentially noxious' industries. 'Frequent' alcohol consumption, on its own, caused a marginally elevated risk for oesophageal cancer (odds ratio=1.7, 95% CI 1.0–2.9, for women and odds ratio=1.8, 95% CI 1.2–2.8, for men). The risks for oesophageal cancer in relation to alcohol consumption increased significantly in male and female smokers (odds ratio=4.7, 95% CI=2.8–7.9 in males and odds ratio=4.8, 95% CI 3.2–6.1 in females). The above results are broadly in line with international findings.