Age standardising data: What does this mean and why does it matter?
Today the ONS has released topic summary data for health, disability and unpaid care. This data has been age standardised, but what does this mean and why have we done it? Helen Colvin, Head of Census Health and Disability Analysis explains more.
Health, disability and unpaid care, are all closely related to the age of a population. In a more elderly population, you would expect poorer health, more disability and more unpaid care.
In our topic summaries, we’re comparing results for 2021 to 2011. However, comparing between time periods or two different geographic areas can be problematic as the population’s characteristics can vary. For example, we know the number of elderly people in the England and Wales population has increased over this period, so would a finding of poorer health in 2021 be directly due to a decline in health, or just reflect that the population has aged?
Similarly, we now know that very good general health has improved between 2011 and 2021, but without standardising the data, the extent of health improvements could be masked by poorer health of the older population.
There’s a similar problem when we compare between geographies. We might find that disability prevalence in one local authority is much lower than another and conclude that there’s a lower proportion of disabled people in the first area. This could be true, but if one of the populations is younger, this finding isn’t necessarily related to better health, just a younger age distribution.
Similarly, two geographical areas may have the same proportion of disabled people for one age group (known as the age specific rate) but will have different overall rates if the age structure of the population is different.
To enable us to make meaningful comparisons of, health, disability and unpaid care outcomes when we compare across age groups, over time or between geographies, we age standardise the data. Age-standardised proportions (ASPs) take into consideration both population size and age-structure, essentially evening them out so that you can compare like with like.
When using ASPs to compare between local authorities, we can say that the higher proportion of disabled people found in Blackpool and Blaenau Gwent is due to a difference in prevalence, rather than, for instance, a difference in the age or size of the population. If you’d like to see more detail on how ASPs (also known as age standardised rates) are calculated, there’s a useful explainer here.
Standardised data is valuable to users who would want to understand differences in one area in comparison with others. Which areas are doing better or worse? Has health improved or got worse over time? This has implications for policy making, holding organisations that spend public money to account, and informing wider public debate.
Of course, there are instances where it is more useful to use non-standardised data. Service providers like GPs, social services and local charities are more likely to need to know about the population in their area and what proportion have health concerns or a disability. This helps in understanding what needs exist in an area and planning service provision to support them. For this reason, we have also published non-standardised data in tables accompanying our topic summaries.
Our health, disability and unpaid care topic summaries all use age standardised data to enable comparisons. Today we’ve seen an improvement in general health between 2011 and 2021, a fall in the proportion of disabled people and a reduction in provision of unpaid care. To explore these findings in more detail, our next publications will look at differences in age and sex outcomes over time and some possible influences on these findings.
If you want to know about health, disability and unpaid care in your area, why not explore our topic summaries, change over time local reports and Census maps.