Disease measures such as death rates, prevalence or incidence vary with age for certain diseases. Most commonly in developed countries, incidence, prevalence and mortality for diseases such as cancer and heart disease increase with increasing age. Other diseases, such as some infections, occur more commonly at other ages.
When conducting an epidemiological investigation, we often want to compare disease measures in different populations, but if these populations have different age structures, then any differences we see may simply be due to the effect of age. In epidemiological terms, age is a confounding factor (see Chapter 6). Standardisation is the technique that allows us to remove the effect of age when comparing disease measures between populations.
- Direct standardisation: the age-specific rates in the population of interest are applied to the proportion of people in each age band in a specified reference population. This method produces a directly age-standardised rate (DASR).
- Indirect standardisation: the observed mortality/morbidity pattern in a population is compared with what would have been expected if the age-specific rates had been the same as in a specified reference population. This method produces a standardised ratio (e.g. SMR, standardised mortality ratio or standardised morbidity ratio). By definition, the SMR for the reference population is 100. Areas, or population groups, with SMRs below 100 have a lower mortality/morbidity than the reference population and those with an SMR of above 100 have a higher mortality/morbidity.
- Standard populations: a national population structure is often used as the reference population, such as that of England and Wales, but standard populations have also been constructed for international comparisons, such as the European and World populations. Both DASRs and SMRs can be calculated using any population as the reference population.
Standardisation is most commonly done for age and sex, but can also be applied to other population characteristics or confounding factors, such as ethnicity or socio-economic status.
Figure 7a shows the number of deaths from cancer in Town A, Town B and England and Wales in 1 year. As the purpose of this example is to demonstrate the technique of standardisation, we have used only six broad age bands; normally, 5-year age bands are used, from 0–4 years to 85 years and more. Also shown are the numbers of people and the age-specific death rates for each age band. Note that the crude all-ages death rates for Town A and Town B are very different, and that the two areas have different age structures, but that the age-specific rates are generally similar.
The DASR is calculated by multiplying each age-specific rate in Town A and Town B by the number in each age band for England and Wales. For example, the age-specific rate at age 35–64 years for Town B is 164.81 per 100,000; multiplying it by the number in that age band for England and Wales gives us 39,246, which is the number of cancer deaths we would expect to see in England and Wales in the 35–64 years age group if the Town B death rate for that group applied across the national population. Summing the numbers across the age bands and dividing by the total population for England and Wales gives us the DASR, which is multiplied by 100,000 to give us a rate per 100,000 population, as shown in Figure 7b. Note that the rates for Town A and Town B are now very similar.
The SMR is calculated according to following formula: