How open-ended age bands can mislead
| 85–89 years mortality rate/y | 90+ years mortality rate/y | 85+ crude mortality rate/y | % 85+ age band over 90 | “Crude” life expectancy years | “Actual” life expectancy years | |
|---|---|---|---|---|---|---|
| Population A | 0.58 | 0.74 | 0.615 | 22% | 1.63 | 1.70 |
| Population B | 0.48 | 0.89 | 0.607 | 31% | 1.65 | 1.37 |
| Rate ratio | 1.20 | 0.83 | 1.008 |
Populations A and B have similar crude mortality rates in the 85+ age band, with slightly smaller “crude” expectation of life from age 85 based on an exponential assumption in population A. However this apparent similarity is misleading and results from the different age structures in the 85+ age band, as shown by the percentage of the 85+ age band who are over 90 years old. In fact, within the 85+ age band the age specific rate ratios are not constant, and when this age band is subdivided the expectation of life from age 85 is actually greater in population A (1.70 years v 1.37 years). Likewise a CMF or SMR based on a single 85+ age band will differ from ones which subdivide the age band.