Author's reply to: Lung cancer mortality in sub‐Saharan Africa

Dear Editor, We thank Becher and Winkler for their comments on our most recent (2008) estimates of the worldwide burden of cancer.1, 2 In their letter, they express their disagreement with the methods and results for the sub-Saharan African region. In particular, they claim that our estimate of lung cancer mortality in the region is too low, and, because these estimates are based on reported cancer registry data, the overall number of cancer deaths in the region is also under-estimated. To support this conclusion, they present a method of modelling lung cancer mortality rates that relies on (i) country-specific estimates of smoking prevalence (by age and sex); (ii) lung cancer mortality rates in nonsmokers and (iii) relative risk estimates for current and ex-smokers. In their comparative study,3 Becher, Winkler and colleagues estimated the smoking prevalence in sub-Saharan Africa using a single survey analysis from Ethiopia, assuming a 25% smoking prevalence, while the most recent WHO assessment in Ethiopia is only 3.3%.4 Therefore, it is difficult to conclude that smoking prevalence in many African countries is similar to that in the South African Republic (an atypical sub-Saharan African country) and that the lung cancer rates should be similar. There is no reference to the estimates of the lung cancer mortality rates in nonsmokers, but we presume that these were the same as those derived by observation and modelling of data from Northern Europe, the United States and Japan,3 and somewhat greater than the rates observed in the CSP-II study in the United States.5 Nothing is known of the mortality rates of African nonsmokers; it seems plausible that they would be lower than in the United States and Europe, where environmental tobacco smoke, other types of air pollution (and possibly dietary factors) will contribute to the risk. The relative risk estimates for tobacco smoking are population-specific (determined by the smoking history) and are not transferable from one population to another, and are certainly very much lower in Africa than in contemporary populations from the Nordic countries and the United States, where current smokers have had a much longer exposure to tobacco. In our GLOBOCAN 2008 paper,1 we recognize that the results for many sub-Saharan African countries (except the South African Republic), based as they are on either sparse local cancer registry data or on data from neighbouring countries, probably underestimate the "true" numbers of lung cancer cases or deaths in the region, especially in women. The same might be said of Indonesia where the GLOBOCAN estimates are based on data from cancer registries in neighbouring countries, without mortality data on lung cancer, which the correspondents erroneously believed contributed to the concordance of their estimate with GLOBOCAN. It seems equally plausible that the modeled estimates developed by Becher and Winkler represent a substantial overestimation because of the lack of information on the smoking prevalence, and the doubtful assumptions concerning the two other parameters. In present day Africa, cancer registration remains the only realistic basis for robust estimates of cancer incidence and mortality. Those presented in GLOBOCAN 2008 are the most accurate that can be made at present and can be used in support of planning for cancer control. We fully recognize that there are limitations in the reliability of cancer registry data not only from sub-Saharan Africa but also in many other developing countries. Promoting cancer registration in these areas of the world is one of the activities of the Cancer Information Section at the International Agency for Research on Cancer. In the meantime, we welcome collaboration with other groups in additional research, including modelling, to supplement the current local data in these regions. Jacques Ferlay*, Hai-Rim Shin*, Freddie Bray*, David Forman*, Colin Mathers , Donald Maxwell Parkin , * Section of Cancer Information, International Agency for Research on Cancer, 150 cours Albert Thomas, 69372 Lyon Cedex 08, France, Department of Measurement and Health Information Systems, WHO, Geneva, Switzerland, Clinical Trial Service Unit & Epidemiological Studies Unit, University of Oxford, Oxford OX3 7LF, United Kingdom.