Counting the dead in China

Papers pp 1411, 1423

Purely descriptive statistics on the numbers of people dying from different causes at various ages can be of enormous importance. But it is not enough merely to count “who dies of what disease”: the causes of those diseases, particularly in large populations, need also to be reliably measured and their evolution monitored. In rich countries the established vital registration systems, in some cases dating back over 100 years,1 can be used to assess disease patterns and trends, while decades of epidemiological research have identified some of the principal causes of such trends, particularly tobacco use. 2 3 In poorer countries, however, vital registration systems are not yet sufficiently well developed to document disease trends and cannot support large scale studies of the avoidable causes of disease.4

To assess the main patterns of mortality and the main avoidable causes of premature death in large developing populations, innovative, robust epidemiological methods are needed, and China has been particularly ingenious inthis respect. This week Liu et al report the world's largest analytical study of tobacco deaths, showing that in China smoking is already causing about 750 000 deaths a year and predicting that this will rise to three milliona year by the time the young smokers of today reach middle and old age (p 1411).5 Also in thisweek's issue, Niu et al confirm these conclusions by a large prospective study of smoking and death (p 1423)6 set in China's nationwide system of disease surveillance points.7 These studies are important both for their concordant conclusions about tobacco and for their epidemiological methods.

During the past decade the establishment of a nationally representative system of 145 disease surveillance points covering only 1% of the Chinese population has yielded reliable statistics on death and disease for the entire country, at relatively low cost. 78 A typical surveillance point coversa defined population of about 100 000, in which a team of workers, including a doctor, investigate each death and assess the underlying cause from medical records and interviews with family members. This system could be extended to other countries and is particularly useful for prospective studies such as that of Niu et al.6

Prospective studies, however, take years to mature, whereas the novel retrospective methods of Liu et al require little more than reliable statistics on cause of death for adult populations plus reports from families of the smoking habits of the deceased. Proportional mortality analyses then reveal the excess of smokers among those dying from neoplastic, respiratory, orvascular diseases, using deaths from other diseases as controls. Thus, for example, the excess of smokers among those dying of lung cancer is used to infer the excess of lung cancer among smokers. (Similar estimates of tobacco attributed mortality are obtained if surviving family members are chosen as controls.)

The validity of such proportional mortality analyses5 is confirmed by the large prospective study,6 which ascribes an identical proportion (12%) of male adult deaths in 1990 to tobacco. This percentage will, however, probably more than double over the next few decades,5 as a delayed effect of the large increase in cigarette consumption among men over the previous few decades.

But, although the overall risk of death may well become about as big forChinese as for Western smokers, the detail is surprisingly different. In China tobacco causes far more deaths from chronic lung disease than from vascular disease (indeed in China smoking causes about as many deaths from tuberculosis as from heart disease), causes widely different lung cancer risks in different Chinese cities, and causes many deaths from cancers of the oesophagus, stomach, and liver.5

Similar retrospective methods could be used to monitor the hazards of tobacco in many other populations where cause of death data can generally be obtained for adults. The routine reporting of smoking status on the new South African death certificate9 is of particular research interest as this might obviate the need for follow up interviews of family members.

These two new studies provide the first nationwide evidence of tobacco's effects in a developing country. The hazards are already substantial and they cannot be limited to China. Worldwide, by the turn of the century, cigarettes will already be causing about 4 million deaths a year, half in rich countries, half in poor countries. But if current smoking patterns persist then by about 2030 this will have risen to 10 million deaths a year, 70% of them in the developing world.10