Changes to health state life expectancy methodology

We Asked

ONS currently produces Health State Life Expectancy estimates for local areas of the UK. These estimates at both national and subnational level are calculated using health state prevalence data available from the Annual Population Survey (APS). Stakeholders have wanted estimates of these summary measures of population health at a subnational level, and they are a national indicator in the Public Health Outcomes Framework.

We tested three alternative methods, designed to address the current weakness of small sample sizes producing somewhat erratic health state prevalence estimates across the age distribution in those areas with smaller populations. Each method modelled age-specific health and disability-free state prevalence using a least squares regression containing a quadratic line of best fit.

ONS wanted to ensure stakeholders had the opportunity to respond to the proposed change in methods, having had the opportunity to appraise its impact in the Proposed method change in UK health state life expectancies paper published in December 2017.

ONS wanted to elicit the opinion of key stakeholders on the proposed change, particularly regarding its complexity and ease of communication. We also asked whether other methods should be considered, whether they had any concerns with us implementing the method and whether implementation would impact on their use of these statistics.

You Said

We are grateful to everyone who took time to respond to the consultation. There were 5 responses received, but some of these were a collation from separate organisations although not all wanted to be identified. Those willing to be identified can be found in the final section of the consultation report titled ‘List of responding organisations’.

The key points to note from the responses to the consultation are:

• Further complexity to improve the model should not be undertaken at the expense of ease of communication of the statistics. It was thought the modelling approach was already somewhat complex and shouldn’t be made more so.

• Four respondents agreed that the method proposed to model health and disability state prevalence reduces the volatility in the estimates for sub-national areas and at the older ages, making comparisons over time easier.

• The proposed method complexity can be simplified by explicit explanation of the changes in methodology in plain English, which will mitigate the risk of figures being misinterpreted.

• Concerns were raised as to whether health and disability questions would be asked in the 2021 Census, and if they did, would they be comparable and would changes present challenges for using the new proposed method. Over longer time horizons, there was also a concern about the adequacy of data expected to replace the Census in 2031.

• ONS must ensure sufficient information is available to allow other departments and organizations to calculate figures and not be reliant on ONS calculating the health state life expectancy estimates using other sources or for different population sub-groups. This will also reduce the risk of users misrepresenting figures.

• There was a concern raised over whether the method will continue to be developed, as this would cause further breaks in the time-series. Another respondent expressed concern that the implementation of the method would put at risk a long, consistent time-series.

• One organisation has said they will implement the method that ONS decides to use, provided sufficient information is available to allow them to replicate the estimates. The proposed changes provide data for all subnational areas which would make the data more usable for one organisation’s purposes.

• Respondents said the health state life expectancy estimates feed into public health frameworks for England and Wales, which help understand how public health is being improved and protected. The estimates also have policy impact in that they are used in consideration of the State Pension age.

We Did

As a result of this consultation the following actions will be undertaken:

• We will implement our proposed method based on modelling health and disability-free state prevalence to calculate future health state life expectancy estimates and provide a back series;

• Our next publication for sub-national areas of the UK, due to be published in December 2018 will apply this method;

• Our publication measuring the socioeconomic inequality in health state life expectancy in England and Wales will be published in March 2019 and will also apply the new method.

• Subsequently, ONS will no longer publish statistics based on the previous method;

• The December 2018 and March 2019 publications referred to above will also take account of the revised mid-year population estimates for sub-national and small area populations in England and Wales and provide a back series from 2009 to 2011 for the former and from 2011 to 2013 for the latter;

• The December publication will include additional areas such as combined authorities (geographic codes E47) for England and Health Boards for Wales;

• The health state life expectancy template will be updated to align with the new method and aid users’ understanding and to help them to be able to calculate their own estimates;

• Will update the Quality and Methodology Information (QMI) report to communicate the method change concisely, the reasons for it, and its impact on the existing time-series;

• Facilitate further work with the devolved administrations to ensure understanding of the methodology to allow other departments to replicate the estimates and calculate their own estimates with other areas of interest;

• We will additionally undertake an evaluation of the method’s performance, comparing the 10-year trajectory between 2010 to 2012 and 2020 to 2021 using the new method and that using the censuses to interpolate census-based prevalence over the decade;

• ONS will ensure any changes brought about by what is learnt from the 2021 Census data will be implemented with a back series. If the 2021 Census contains health questions, ONS will wait until a formal evaluation and interpolation study has been conducted and published on. This is unlikely to be published until the late autumn of 2023.