This short story explains a change in statistical methods this year which will increase age-standardised mortality rates and cancer incidence rates published by ONS. Previous figures will be updated so that comparisons over time can still be made. This is a technical improvement following the updating of a long-standing international standard method – there has been no actual change in the numbers of deaths or cancer cases reported by ONS. A consultation in autumn 2013 showed that 21 of 22 experts responding supported ONS’s approach to implementing the change, which follows a European initiative. Further details are available in technical papers previously published on the ONS website.
What is the European Standard Population?
The European Standard Population (ESP) is an artificial population structure which is used in the weighting of mortality or incidence data to produce age-standardised rates. Eurostat, the statistical office of the European Union, has decided to update the structure to be more representative of the current population of Europe (Eurostat, 2013).
The ESP was first introduced in 1976 to facilitate comparison of disease rates between populations of different average ages. Until 2013, this ESP structure had not been altered since its initial implementation. Eurostat recognised that the 1976 ESP is notably younger than the actual population in the last decade, and so its use does not adequately reflect the current population structure. As a result, a new ESP structure (2013 ESP) has been implemented; this allocates a greater weight to the older population to better reflect the ageing population (Population of England and Wales, 1911-2011).
The implications for methodology
The ESP has become an accepted methodological standard in health statistics in the UK and the rest of Europe, and is used in the calculation of age-standardised rates by ONS, government departments, academic research and the National Health Service. It is important to consider the impact of the new ESP on age-standardised mortality and morbidity rates (Office for National Statistics, 2013).
The 1976 ESP has an upper age band of 85+ years (see Age-standardised mortality rate calculation template, 2012 (64 Kb Excel sheet) ), whereas the 2013 ESP contains age bands of 85-89, 90-94 and 95+. The 2013 ESP gives older ages a greater weighting than the 1976 ESP (see Age-standardised mortality rate calculation template, 2013). It is important to note that population data, particularly at sub-national levels, are often produced using specific upper age limits (often 85+ or 90+). Therefore, analysis here and for National Statistics outputs in the future will be based on the upper age bands: 80-84, 84-89 and 90+.
Comparison of rates between 1976 ESP and 2013 ESP for England and Wales
The 2013 ESP mortality rate is considerably greater than the 1976 ESP rate for conditions which are commonly associated with older ages. This can be seen by looking at deaths from dementia (including Alzheimer’s) in 2012. Using the 1976 ESP, the mortality rate was 33.2 deaths per 100,000 men and 39.5 deaths per 100,000 women. In contrast, using the 2013 ESP 90+ and based on the same year of data, the mortality rate was 80.7 deaths per 100,000 men and 93.7 deaths per 100,000 women. This amounts to a rate difference of 47.5 for males and 54.2 for females.
The significant difference seen between these two ESPs can be explained by the nature of this disease. In 2012, 96% of men and 98% of women who died of dementia (including Alzheimer’s) were aged 70 or over. As the 2013 ESP has an older population structure than the 1976 ESP, there is a significant impact on the age-standardised rates calculated for those causes of death.
In comparison, in 2012, only 16% of males and 34% of females who died from land transport accidents were 70 or over. Using the 1976 ESP, the mortality rate was 4.2 deaths per 100,000 males and 1.2 deaths per 100,000 females. In contrast, using the 2013 ESP 90+, the mortality rate was 4.5 deaths per 100,000 and 1.4 deaths per 100,000 respectively. This illustrates how the 2013 ESP has less impact on mortality rates for conditions where death counts are not clustered in older age bands.
The majority of age-standardised mortality rates by ICD chapter in England and Wales increase using the 2013 ESP
For age-standardised mortality rates by chapter of the International Statistical Classification of Diseases (ICD 10th Revision, 2014), all age-standardised rates stayed the same or increased using the 2013 ESP in comparison to the 1976 ESP. The only exception is rates for certain conditions originating in the perinatal period (chapter 16), where the rate decreased by 0.5 deaths per 100,000 males and 0.4 deaths per 100,000 females using the 1976 ESP, to 0.3 deaths per 100,000 males and 0.3 deaths per 100,000 females using the 2013 ESP. This is because these deaths are most likely to affect infants. For example, in 2012, 93% of the deaths from these causes were babies under one-year-old. Rates for these causes would generally not be age-standardised, but would be presented as age specific rates using live births as the denominator.
Where can I found out more about mortality statistics?
These statistics were compiled and analysed by the Life Events and Population Sources Division. If you’d like to find out more about our mortality statistics or the implementation of the European Standard Population, visit our Health and Social Care theme page. A dedicated page has been set up with technical papers about the ESP change. If you have any comments or suggestions, we’d like to hear them. Please email us at: email@example.com.
Office for National Statistics. Revised European Standard Population 2013 (2013 ESP)
Office for National Statistics. Age-standardised mortality rate calculation template, 1976 ESP (64 Kb Excel sheet)
Office for National Statistics. Age-standardised mortality rate calculation template 2013 ESP (93.5 Kb Excel sheet)
Office for National Statistics. 100 years of Census, England and Wales Population 1911- 2011, 2013
World Health Organisation, International Statistical Classification of Diseases and Related Health Problems 10th Revision