| Popis: |
Using the simulation model context we checked the consistency of incidence, prevalence and mortality data for specific types of cancer so as to construct an appropriate data set for chronic disease modelling. The data sources used were IKZ for regional cancer prevalence rates and survival proportions, NKR for national incidence rates, and CBS for national registered cause-specific mortality rates. Consistency checks were made for lung, rectal, colon, stomach, oesophagus, breast and prostate cancer. Using national incidence rates and regional survival proportions and a two-state transition model, we calculated prevalence and population mortality rates. We compared the calculated rates to empirical ones. For most types of cancer the calculated prevalence rates fitted well to the empirical ones, except for lung cancer (females), breast and prostate cancer. We found several possible explanations for these differences. The first one was double counts in case of multiple tumours. Based on data we corrected for double counts for breast cancer. The second explanation was differences between regional and national morbidity figures. This was confirmed for lung cancer (females) and prostate cancer. The third explanation was past trends in disease incidence. Correcting for these trends resulted in smaller differences for lung cancer (females), breast and prostate cancer. The calculated mortality rates fitted well to the empirical ones for the lethal types of cancer, but less well for those types of cancer with better prognosis, namely rectal, breast and prostate cancer. These differences could be explained by competing death risks. Several studies have confirmed this explanation. |
