Abstract
A method is presented for projecting mortality rates for certain causes on the basis of observed rates during past years. This method arose from a study of trends in age‐specific mortality rates for respiratory cancers, and for heuristic purposes it is shown how the method can be developed from certain theories of cancer induction. However, the method is applicable in the more common situation in which the underlying physical processes cannot be modeled with any confidence but the mortality rates are approximable over short time intervals by functions of the form a exp bt, where b may vary in a continuous, predictable fashion as the time interval is varied. It appears from applications to historical data that this projection method is in some cases a substantial improvement over conventional curve‐fitting methods and often uncovers trends which are not apparent from observed data.
Original language | English |
---|---|
Pages (from-to) | 205-215 |
Number of pages | 11 |
Journal | Risk Analysis |
Volume | 1 |
Issue number | 3 |
DOIs | |
State | Published - Sep 1981 |
Keywords
- forecasting
- mortality rates
- relative risk
- respiratory cancer
- trending