Methodological problems in comparing incidence and prevalence of leukaemias and lymphomas: ascertainment and age adjustment.

For a reliable incidence of a disease, all the cases that occurred in the area during the period of interest should be diagnosed, registered and classified correctly. It is very important to know in comparisons which data sources a particular cancer registry is utilizing and for how long it is waited until a publication of incidence is released. Examples from Finland show that deficits of up to 5-25% are possible in leukaemia and multiple myeloma. The populations to be compared may also have large differences in the age distribution. As the risk of all cancer, also for leukaemia and lymphomas, increases strongly by age, it is crucial that care is taken about the age adjustment before showing the incidence rates for comparisons. The age adjustment can be done using different methods and standards. In practice, the choice of methods for comparative purposes is often not crucial. It is also essential to evaluate the age-specific rates when making comparisons between populations. For cancer prevalence it would be important to know which patients can be considered cured. A population-based cancer registry does not, as a rule, have this information. However, if a cancer registry follows up the patients, survival rates of the patients may, after a number of follow-up years, start to resemble those in a comparable general population group. This is an indication that the living patients are cured and should be deducted from cancer prevalence. Advances would also be needed in statistical methods and improvement in the quality and coverage of cancer registration.