Relative value of incidence and mortality data in cancer research.

Cancer surveillance has played an important role in programmes of cancer control, ranging from aiding formulation of current hypotheses regarding the nature of the causes of cancer to assessing the effectiveness of cancer treatment regimes and cancer prevention programmes. Central to this has been the contribution from routine data collection schemes, particularly cancer mortality data and cancer registration schemes, the latter providing cancer incidence statistics for a variety of international populations. Criticisms have been made of the quality of cancer incidence data and there have been suggestions that cancer surveillance may be better achieved by use of mortality data. From examination of the reliability and quality of mortality data, it would appear that international variation in the quality of death certification and in the application of internationally agreed rules for selecting the underlying cause of death may in themselves be enough to vitiate the argument that there is significant international variation in cancer levels or to indicate variation where none in reality may exist. Good cancer incidence data are vitally important to descriptive epidemiology as are good cancer mortality data. It is important to recognize that there are limitations to both types of data which vary both temporarily and internationally. Cancer surveillance and the assessment of the impact of cancer control programmes depend on the reliability of descriptive epidemiology and would best be achieved by maintaining current, population-based cancer registration schemes and, if and where possible, extending such schemes to other populations or population groups. Maximum benefit would be achieved by simultaneous improvement in the quality of mortality data.