Review of disability data harmonised standards

Policy details

Metadata item Details
Publication date:28 March 2023
Owner:GSS Harmonisation Team
Who this is for:Users and producers of statistics
Type:Harmonisation standards and guidance

The Government Statistical Service (GSS) Harmonisation team are based in the Office for National Statistics (ONS). The team supports harmonisation of data across the UK. This involves providing bespoke advice and harmonised standards and guidance about how data on a range of different topic areas are collected and presented. The standards and guidance are designed to improve the consistency, coherence and comparability of statistics.

Producers of statistics use these harmonised standards as a starting point in their data collection process. By using harmonised standards, they can:

Harmonised standards also allow people to effectively and accurately compare data that has been collected across different datasets. This means we can more easily understand what that data does, and does not, tell us. This ensures that statistics are used to their full effect for the public good.


Measuring disability can be particularly difficult. Many conditions change over time and some people have more than one condition at the same time. There are also different ways to consider disability. For example, we could consider disability according to legal definitions, medical conditions, or societal barriers. Furthermore, health is a devolved issue and different administrations take different approaches to disability.

This review of the disability harmonised standard fulfils a commitment published in our February 2022 GSS Harmonisation Team Workplan. We pledged to review and research the topic area with the aim to ensure that disability standards meet user needs, and to explore any potential improvements.

As part of this review we:

  • worked with a variety of stakeholders, including respondents, interviewers, data processors and data users to establish individual needs and gather feedback on whether the current disability standards could be improved
  • did a survey with data collectors and processors to determine technical capabilities
  • reviewed several international approaches for measuring disability
  • did an unmoderated card sort to investigate potential impairment groupings

The information in our disability and mental health reviews was researched in different stages. The first stage of the project was a desk review of freely available data, including Census 2011, question development reports, and articles in Health Statistics Quarterly. The second stage involved gathering needs from data users, collectors, producers, and providers. Our aims were to:

  • consult with stakeholders across government and devolved administrations about needs
  • engage with teams and departments who are currently doing research in this area
  • define concepts to be measured, consider IDTF recommendations and propose these to stakeholders
  • review and document existing disability and mental health questions used elsewhere in government and internationally
  • recruit members of the public with lived experience

We split stakeholders into four groups to help us consider how the standard meets the needs of a range of stakeholders. These groups were:

  • data users – people who use collected data, for example policy makers or charities
  • data providers – people whose data is captured, for example users of a service that collects administrative data or survey respondents
  • data collectors – people who programme and run data collections, for example survey owners
  • data producers – people who take collected data, clean it, and validate it, for example statistical teams

We understand there may sometimes be overlap between group membership, and these groupings could be seen as artificial. But the goal of this segmentation was to ensure we worked with a range of stakeholder types to find out about their needs, rather than working only with data users. These groups are different to the Office for National Statistics (ONS) user personas, as those personas are for people who use the ONS website. Everyone who uses the ONS website would fall under the “data users” category.

Data users

Our aim for data users was to represent a wide variety of organisations, including policy makers and charities. We wanted to understand what decisions and activities are informed by harmonised standards, which would help us work out what data is needed. Our main work included interviews and workshops with a variety of different data users from both government and non-government departments.

We arranged both informal and formal interviews and workshops with a wide range of different stakeholders. These sessions were structured by a topic guide which prompted questions on the disability harmonised standards and mental health needs. You can find more information on needs for mental health data in the mental health publication. We covered topics such as:

  • their views on the current harmonised standards and IDTF recommendations
  • how they use disability and mental health data
  • what data they would like to have around the topics of disability and mental health

We conducted thematic analysis to identify the main themes and form need statements.

We then held group sessions to review the need statements. These were used to provide evidence about the issues raised by stakeholders and the suggestions they made about the harmonised standards. We grouped over 200 stakeholder need statements into sub-level themes. We then produced a list of common high-level themes which we used to identify our main stakeholder needs.

Data providers

We wanted to speak to data providers to understand their mental models and the language they use to describe disability and mental health. This involved a few different research streams and activities, including:

  • card sorts
  • interviews
  • qualitative surveys
  • working with disability networks

We used this work and information from different user types to assess whether the current standards could be improved.

Firstly, during the data user research we spoke to some individuals who were disabled and provided views from a data provider’s perspective. We also considered findings from cognitive interviews aimed at testing questions about access to services and disability. These interviews were done in January 2022 and involved testing variations of the long-lasting health conditions and illness, activity restriction and impairment harmonised standards with data providers. It is important to note that the long-lasting health conditions and illnesses question was presented as a two-part question instead of a single question and only the first question in the activity restriction standard was used. The interview responses were analysed using thematic analysis and this information identified main themes which have informed user needs statements.

We also worked with Disabled People’s Organisations, gathered relevant findings from disabled people in the interviewer survey, and gathered evidence from the civil service disability networks, including the neurodiversity network. We conducted thematic analysis of the results of these engagement activities to form need statements.

Data collectors and producers

We wanted to understand technical capabilities and limitations for data collectors and producers in using harmonised standards. The needs of these user groups were captured through a survey of collection and processing capabilities.

The GSS Harmonisation Team did a survey to identify the data collection and processing capabilities of:

  • government
  • devolved administrations
  • private sector data collectors and data producers

These surveys were run between March and June 2022. They included stakeholder mapping to identify people that use harmonised standards.

Wave one took place between March and April and concentrated on contacts identified through a stakeholder mapping exercise, a harmonisation champion network, and in the implementation review. Wave two took place in June and concentrated on producers of official statistics and any additional contacts identified in wave one. This activity aimed to explore capabilities of data collectors and data processors, concentrating specifically on capabilities around implementing changes.

A total of 43 data collectors and data processors took part in our survey. We achieved a good representation of respondents based on where they work, the type of data they work with, and modes of collection. The data was analysed to form needs statements.

Data collectors and producers also raised issues during our research that were relevant to the content of the disability standards.

International approaches

It is important to consider international approaches to measuring disability. We researched the approaches used by international organisations within English-speaking countries. These approaches were reviewed for their overall suitability in the UK and for aspects that could be useful for the future development of disability harmonised standards. The organisations and countries researched are:

  • International Classification of Functioning, Disability and Health (ICF) and World Health Organisation’s Disability Assessment Schedule (WHODAS)
  • Washington Group (WG)
  • Global Activity Limitation Indicator (GALI) – European Union (EU) approach
  • United States of America
  • Canada
  • New Zealand
  • Australia

International Classification of Functioning, Disability and Health (ICF) and World Health Organisation’s Disability Assessment Schedule (WHODAS)

The International Classification of Functioning, Disability and Health (ICF) is the World Health Organisation’s (WHO) framework for measuring health and disability. It measures at both individual and population levels and considers how people function in an environmental context.

The ICF presents a biopsychosocial model where disability and functioning are outcomes of interactions between health conditions, including diseases, disorders and injuries, and contextual factors. But the ICF is considered impractical by the WHO for assessing and measuring disability in daily practice. They developed the World Health Organisation Disability Assessment Schedule (WHODAS) as a solution to this.

WHODAS is a generic assessment tool for health and disability, which can be used at population level or in clinical practice. WHODAS aims to reflect the core features of the ICF and is designed to assess people’s experiences with limitations on their activity and restrictions on how they can participate in society regardless of medical diagnosis. There are variations on the question set. The longest version contains 36 questions and the shortest version contains 12 questions. The questions cover the following domains of functioning:

  • Cognition – understanding and communicating
  • Mobility – moving and getting around
  • Self-care – hygiene, dressing, eating, and staying alone
  • Getting along – interacting with other people
  • Life activities – domestic responsibilities, leisure, work, and school
  • Participation – joining in with community activities

WHODAS also provides definitions for all the terminology they use. This includes a definition for impairment which they say means “Loss or abnormality in body structure or physiological function (including mental functions)”. It is important to note that “Abnormality” in this definition strictly refers to a significant variation from established statistical norms and should be used only in this sense.

Use in the UK

WHODAS would not be suitable for UK stakeholders because of the length of the question sets and the costs associated with this. WHODAS does not align with UK legislation. It asks participants to consider their experiences in the last 30 days, compared to the 12-month time frame stated in the Equality Act (2010) and the Disability Act (1995). But there are useful insights to be gained from the language definitions and guidance within the question sets. This insight will be used when we:

  • explore the use of the term “disabled”
  • consider additional guidance on “a little” and “a lot” response options
  • design and test changes for the current UK harmonised standards

Washington Group

The Washington Group on Disability Statistics (WG) aims to promote and coordinate international cooperation on health statistics. The main goal of the WG is to provide comparable disability data internationally. They provide 6 question sets and translations to do this.

Our research has concentrated on the “WG short set on functioning” (WG-SS) and the “WG short set on functioning – Enhanced” (WG-SS-Enhanced). This is because there is little demand for more questions in the UK. The questions aim to get a respondent to concentrate on difficulties they may have doing basic activities. They are asked if they have no difficulty, some difficulty, a lot of difficulty or whether they cannot do something at all, even if they have an aid, such as glasses, or a hearing aid. The questions relate to the following six core functional domains:

  • Vision
  • Hearing
  • Mobility
  • Cognition, or remembering
  • Self-care
  • Communication

The WG-SS-Enhanced adds two further domains of functioning:

  • Upper body activities
  • Affect — which includes Depression and Anxiety

The WG-SS-Enhanced asks 12 questions covering these 8 domains. The six questions from the WG-SS are embedded within those 12 questions.

Use in the UK

Previous ONS research compared UK approaches with the WG-SS and the Washington group short set on functioning – Enhanced (WG-SS- Enhanced). This research concluded that the WG-SS does not align with the Equality Act (2010). This is because the scope of the WG-SS was too narrow compared to UK legislation. For example, the WG-SS does not include mental health issues. The UK has formally said that they support the use of the questions where countries do not have their own question sets developed, but they do not match the policy definition we need for the UK.

The research also found that the question sets identify substantially different but overlapping groups of disabled people, which can result in a range of different estimates of disability prevalence. The findings conclude that neither approach is better than the other, but the Equality Act (2010) definition is essential in the UK to meet needs of government policy and equalities monitoring. The research notes that the Washington Group questions are considered best practice internationally.

In our future research we will consider the language and questions in the WG question sets to help us design guidance for the UK harmonised standards to meet the Equality Act (2010) definition. The WG questions sets could be particularly useful when we consider examples or guidance for the impairment harmonised standard.

Global Activity Limitation Indicator (GALI) – EU Approach

The Global Activity Limitation Indicator (GALI) exists to fulfil a need for increased EU harmonised data on health and disability. It was proposed for use alongside the Self Perceived Health Variable in various European Social Surveys. The GALI approach considers the concept of “functional limitations” which includes a minimum of four to six variables, including difficulties in seeing, hearing, walking, cognition, self-care, and communication. GALI uses a shorter approach because the WG questions can be difficult to use as a result of extra variables.

GALI is a single-item survey question which has been designed to be easy to translate. It can be used to target situations where health disorders and conditions are affecting peoples’ usual activities. The GALI question is:

“For at least the past six months, to what extent have you been limited because of a health problem in activities people usually do?”

The response options are:

  • “severely limited”
  • “limited but not severely”
  • “not limited at all”
Use in the UK

GALI is criticised for being culturally biased and unreliable when different collection methods are applied. GALI also refers to the past six months, which means it does not align to the 12-month requirement in UK legislation.

We already know people feel the UK harmonised standards are very subjective. People also have concerns around how reliable and consistent responses to the UK harmonised questions can be. We think there would be the same problems with GALI if it was used in the UK. For these reasons, GALI would not provide a better approach than the current UK standard. This is despite the fact it would allow for comparison across European Countries.

United States of America

The USA’s definition of disability is currently defined by the Americans with Disabilities Act (ADA), as a “physical or mental impairment that substantially limits one or more major life activities”. This includes people who have a record of these kinds of impairment, even if they do not currently have a disability. The effect of a condition on major life activities is assessed without the use of assistive devices. It is also assessed without any actions taken to ease symptoms or improve how a person functions.

The ADA definition is a legal definition and other definitions are used within the US government to assess disability for different things. The Social Security Act definition is used to assess disability for Social Security Disability Insurance. Under the Social Security Act, a person can only receive disability benefits if they are unable to work because of a severe medical condition that has lasted, or is expected to last, at least one year, or expected to result in death. The person’s medical condition must prevent them from doing the work that they did in the past and prevent them from adjusting to other work. They must provide evidence from a doctor of their condition to qualify.

Use in the UK

The various approaches to defining disability in the USA align to different models of disability. The one used to assess for Social Security Disability Insurance aligns to the medical model of disability. In the UK we have been asked to align to the social or biopsychosocial models, which make the US approach unsuitable for use in the UK. But the legal definition used by the ADA contains language that could provide valuable insight when we consider updated guidance or changes to the current UK harmonised standards.


The Canadian government does not have an official definition of disability. They currently use the WHO and United Nations (UN) definitions. “Disabilities” is considered an umbrella term covering impairments, activity limitations, and participation restrictions. People must have at least one marked restriction from the following list, or a combination of two or more if one impairment does not meet the criteria for “marked” restriction:

  • Walking
  • Mental functions
  • Dressing
  • Feeding
  • Eliminating – which refers to bowel or bladder functions
  • Hearing
  • Speaking
  • Vision
  • Life-sustaining therapy

To be considered a “marked” restriction an individual must be unable to do the activity, or take three times longer than someone of a similar age without the impairment, regardless of assistive devices or medication. The restriction must also be present at least 90% of the time. Finally, the restriction must have lasted, or be expected to last, for a continuous period of at least 12 months.

Use in the UK

The Canadian approach to measuring disability does not align with UK legislation. UK legislation covers people who have any illness or health condition that is expected to last 12 months and affects day-to-day activities, regardless of impairment type. But the definition of “marked” restriction has provided useful insight which will be considered when reviewing the response options for activity restriction of “a little” and “a lot”.

We will also research the inclusion of “eliminating” when we research the potential of including continence as a response option in the impairment harmonised standard.

New Zealand

According to the New Zealand Disability strategy for 2016 to 2026, disability is something that happens when people with impairments face barriers in society. This aligns with the social model of disability. In 2018, New Zealand changed their approach to gathering disability data in their census by using the Washington Group Short Set on functioning (WG-SS).

Use in the UK

This social model approach aligns with Scottish and Welsh government preference for the UK harmonised standards to align with the social model of disability. There are conflicting views on which model the UK harmonised standard aligns to, but the devolved government preference for the social model of disability means we will research the New Zealand approach when we consider future alignment to the social model.


Most data collections in Australia identify disability using the concepts described by the WHO in the ICF. Australian data collections also explain that the definition and method of identification of disability depends on the collection type and purpose.

The Australian government recognises that people experience different degrees of impairment, activity limitation and participation restriction. They also recognise that this can be based on several individual or combining factors, both personal and environmental.

One definition of disability used for employment purposes is based on the Australian Bureau of Statistic’s survey of disability, ageing, and carers. People are considered to have a disability if they have a limitation, restriction or impairment which has lasted, or is likely to last, for at least 6 months and restricts everyday activities. This survey is considered the highest standard for disability identification in Australia. It uses a comprehensive question set which includes asking about the level of limitation experienced by disabled people. Respondents are asked to choose from a ranked list ranging from schooling or employment restriction only, to severe or profound limitation.

Use in the UK

This approach demonstrates a combination of two questions in one. Our research ruled out using this question approach because of respondent preference. But the response options which ask about the level of limitation experienced may be useful when we consider further guidance for the response options of “a little” and “a lot” which relate to activity restriction.

Other International Approaches

We did not consider other international approaches from non-English speaking countries as part of this research. This is because of difficulties with translation which may influence the meaning of the questions.

Our findings will be used to inform our research and to update our questions accordingly. Until our updated standards are published, we encourage users to continue to use the current harmonised standards.

The current harmonised standard

One of the most common data needs is to define disability according to its legal definition. There are two relevant harmonised standards that need to be used together to collect data on disability in line with the Equality Act (2010) for Great Britain, or the Disability Discrimination Act (1995) for Northern Ireland:

These standards have been designed specifically to capture the core disabled population according to UK legislation. Users can also collect data using the impairment harmonised standard to understand the activities that a person can and cannot do because of a health condition. There is also harmonisation guidance for mental health data, which explains the context of statistics without providing a preferred standard. We have published a review of the mental health standard alongside this review.

Reasons for this review

The current disability standards were published in 2011 and the question design dates back to 2008, as mentioned in Health Statistics Quarterly. A review of the current standards is now needed, and it was recently recommended by the Inclusive Data Taskforce (IDTF) that standards should be reviewed every five years. We also needed to review the disability standards in relation to the mode of collection and data requirements.

Mode of data collection

Question design began in 2008 with a group of topic experts who created the questions before they underwent cognitive testing, during which they explored whether the questions were performing as expected. The wording and format of the questions were based on a delivery mode in which the questions were asked by a survey interviewer, rather than being completed directly by respondents. Since then, the mode of data collection has changed to include and encourage the use of online self-completion. Since 2008 there has also been more interest in administrative, or “admin”, data-based statistics. Admin data often uses self-completion or proxy completion, and the disability standards were not designed for these modes of collection. Admin data sources are a useful resource to complement survey research. They are inexpensive, sustainable, and timely. But admin data sources are designed for operational purposes and not for research. This means they are unlikely to meet the harmonised standards of consistency, comparability, and coherence. For example, some data sets focus solely on the legal definition of disability, others consider disability from a purely medical point of view and others reduce disability to a single self-reported tick box. The Harmonisation team is dedicated to proactively working with and consulting stakeholders on the advantages of using harmonised standards, such as our ongoing work with NHS England. We are seeking to harmonise admin data from design through to data coding, formatting, and presentation.

As the mode of data collection is changing, there is a need to review how we optimise the standard for the mode and update it as needed. As set out in our upcoming plans, we intend to research and update our standards, firstly considering an online mode. In the long-term, we will then review our questions and optimise them according to telephone and face-to-face modes. We also plan to work closely with data collectors and processors to develop an offering for harmonised admin data.

Data Requirements

Data requirements for disability standards have changed in three major ways over the last 10 years.

Firstly, there is an increasing argument for viewing disability in line with the social model of disability, rather than viewing it according to legislation. Both the Scottish Government and Welsh Government have signed-up to the social model of disability. Existing standards were designed to align with the Equality Act (2010) and Disability Discrimination Act (1995). The Inclusive Data Taskforce (IDTF) put out recommendations in 2021 which included the need to revise measures of disability to make them closer to the International Classification of Functioning, Disability and Health (ICF) biopsychosocial model. The social model stands in opposition to the medical model and places the disablement process on societal barriers rather than in the individual. The biopsychosocial model considers environmental factors but also includes the role of health conditions such as diseases, disorders, and injuries. Questions relevant to these models focus on activity limitations and participation restrictions regardless of medical diagnosis. It is important to review and understand whether we should prioritise aligning our disability standards with legislation or whether this is now an outdated approach.

Secondly, now that the UK has left the EU we need to review whether standards should remain compliant with the European Union Statistics on Income and Living Conditions (EU-SILC) needs. The main difference is that EU data includes health conditions lasting for at least six months, whereas UK legislation includes health conditions lasting at least 12 months.

Finally, there are data gaps where complementary surveys that captured additional progressive health conditions and other non-core elements of the disability standard have either:

  • stopped running – for example, the Life Opportunities Survey
  • changed over time – for example, the Family Resources Survey

Strengths and weaknesses of the current standard

We have identified areas where the existing disability standards may not fully meet needs. But we have also found aspects of the current standards that are performing well. In some instances, the same aspect of the standards was seen as both an issue and a strength depending on the users’ point of view.

Known weaknesses of the current standard

Defining “a little” and “a lot”

Our research showed that respondents do not consistently interpret the difference between the two response options “a little” and “a lot”. In the activity restriction harmonised standard, respondents are asked “do any of your conditions or illnesses reduce your ability to carry out day-to-day activities?”. They can answer this question with one of the following response options:

  • “yes, a lot”
  • “yes, a little”
  • “not at all”

Cognitive interviews showed some respondents felt that the difference between “a little” and “a lot” was subjective. They said this affected their confidence in their answer. One respondent said, “I don’t really understand the difference between a lot and a little”, which demonstrates the point. This finding was supported by Census 2021 for England and Wales question testing, where some respondents wanted an explanation about what counted as “a little” and what would be “a lot”.

This finding was also supported in our 2022 survey of Office for National Statistics interviewers. Some interviewers expressed that the response options are too subjective. One interviewer said, “one person’s day to day activities may be very different to another’s, and therefore the reduction in ability to carry out may be very different”. Interviewers also shared that respondents often asked for clarification on the difference between “a little” and “a lot”.

This issue does not affect a binary disabled or not disabled distinction because the “a little” and “a lot” options are aggregated in line with the legal definition to demonstrate a restriction to daily activities. But there is a user need for information about how severe a person’s condition may be. For example, our stakeholders have reported the need to disaggregate by severity and impairment type to understand how contexts impact disabilities differently. Producers of statistics should include a warning alongside this data to advise users that it is based on a subjective assessment. We will address this in the short-term by publishing a warning alongside the harmonised disability standard. In the long-term we plan to explore the possibility of providing appropriate guidance for defining “a little” and “a lot”. For example, Canada defines a “marked restriction” according to the time it takes a person to complete a task and how often the restriction is present.

Impairments and their relationship with medical conditions

The harmonised standard for impairments presents 10 categories of impairment type and asks users to report whether they are affected by any of the listed impairments. The question does not capture medical conditions and we do not believe respondents or data users fully understand this. The impairment standard is designed to look at the activities a person cannot do, or the activities they have difficulty doing because of their health condition. It does not aim to capture a medical condition. For example, glaucoma is a medical condition, but being unable to see or being partially sighted is an impairment. The relationships between impairments and medical conditions are not straightforward, and often involve highly complex interactions.

Qualitative research has shown that each impairment can relate to multiple health conditions and that each primary impairment can lead to impairments in many other areas. For example, a person who experiences a condition which impairs vision may report other impairments caused by vision loss. One respondent said, “with my vision it causes a lot of distress sometimes and fatigue… if someone is looking to me and talking to me I don’t know if they are looking at me or not. And in that case, there are things that are kind of socially different”. In this specific situation, the respondent expressed feeling unsure about whether they had answered the question correctly, or if they should only have selected the primary impairment rather than all ways they are affected. They said, “I got a bit confused because you kind of have secondary elements to a condition”.

This highlights two issues. Firstly, we cannot be sure whether respondents are consistently answering the impairment question by giving information about both the primary impairment and any secondary impairments, as the question was intended. For example, if a respondent has a primary impairment that affects their vision but also has impairments involving their “mental health”, or “stamina or breathing or fatigue”, these should all be included in their answer.

Secondly, we cannot be sure whether data users are correctly interpreting this data. For example, social or behavioural impairments can be reported for a variety of reasons. But data users have used this response option to indicate autism spectrum disorder. This is an issue which is likely made worse by the response option guidance which states “for example associated with autism spectrum disorder (ASD) which includes Asperger’s, or attention deficit hyperactivity disorder (ADHD)”. Our research participants with academic knowledge of neurodiversity expressed that this guidance implied a diagnosis of ADHD and ASD. They also stated that this meant “their behaviour is going to be poor” and said “you’re inferring that ADHD is linked to behavioural issues and I don’t think that is true”. In the medium-term we will address this finding by amending and testing the guidance for the impairment question and we will specifically address the response option for “socially or behaviourally”.

Incomplete impairment response options

A second issue with the impairment standard response options is the incomplete list of impairment types. The list of impairments included in the harmonised standard was created to “reflect those impairments deemed most useful and relevant for categorisation to aid policy development, policy monitoring and service needs assessment” as described in Health Statistics Quarterly. It was not intended to be a comprehensive list of impairments.

Although the list of impairments was not intended to be comprehensive, recent cognitive testing found the response options largely covered what participants expected. One research participant suggested that we should add a response option for incontinence because this can be a barrier to participating in society. The Health Statistics Quarterly articles note that this was originally considered as a response option because it is included in the list of capacities in the Disability Discrimination Act (1995). But the response option was no longer included by the time the standard reached the cognitive testing phase of development. We may hypothesise that this decision was made because it was not well received, by either respondents or survey interviewers, or because the response option would not have helped with developing policy. We would need to research the decision more thoroughly to understand the reasons why the response option was removed.

We were also asked why “communication” was not included. This response option was also considered when the question was developed. Notes from the original cognitive testing in 2009 reported “when [the respondent] was asked if there were any areas of life, not listed, where he would say he did have difficulty. He said “communication” was such an area”. This shows we may need to do some research to better understand why “communication” was removed as a response option.

We plan to investigate and understand whether “continence” and “communication” should be included as response options in the impairment standard as part of our work in the short-term.

Confusing and incomplete guidance for impairment response options

Another issue with the impairment response options is the incomplete or confusing guidance that is offered. When participants were asked to give ideas about where the standard could be improved, they spoke about the guidance in the response options.

Respondents said they found the guidance confusing. For example, the guidance for “dexterity” gives the examples of “lifting and carrying objects [and] using a keyboard”. A participant who selected this response option said that the guidance did not match their own definition of dexterity. The participant said, “I wouldn’t necessarily think of the term dexterity with lifting and carrying objects… I can lift things, but when it comes to typing I can’t because the skills are different”. Other respondents who did not select this response option said similar things. One respondent said, “lifting and carrying objects is not something I would personally have thought would come under this classification”.

Some response options do not offer any guidance, for example, the “mental health” response option. Participants commented on the uneven use of examples, and they expressed interest in the idea of providing examples for all response options. One respondent said, “I appreciated the little explanations here, maybe with mental health there could have been some clarification there”. Some data users also noted that “mental health” was a broad category and that the lack of guidance made them unsure about the information that was or was not being captured.

Our discussions with stakeholders have also identified a potential need to change the response option of “mental health”. Data users said they would like more detail about mental health because of the amount of response options for physical impairments and lack of comparable detail for mental impairments. But data collectors and data users were also aware that adding more response options may make each category too small to analyse.

We aim to address these issues in the medium-term. We will design and test guidance examples for all response options that do not currently have any guidance attached to them. We will also research improvements in the examples that respondents found confusing.

Impairment option sample sizes

The impairment response options were selected based on how useful and relevant they were in helping to develop policy. Although it is not a comprehensive list of impairments, the number of options still means that sample sizes are often too small for analysis. The Office for Statistics Regulation (OSR) noted in their Review of Transport Accessibility Statistics that “where impairment specific data are collected, they are often not published because sample sizes are too small to provide breakdowns for each of the 10 impairments included in the harmonised standard”. This led to their recommendation that “the GSS Harmonisation Team should develop high level grouping to enable publication of data that represent different experiences – for example high level barrier groupings, and/or high-level impairment groupings”. While we have been working with data users and data processors, we have found the issue of small samples and inability to provide breakdowns is wider than just transport statistics. The same problem exists across both the public and private sectors.

To remedy this, OSR proposed two potential groupings to solve this problem:

  • “visible/invisible impairment”
  • “mobility/cognitive/sensory”

To understand whether these are the most appropriate groupings we conducted a card sorting exercise with over 1000 members of the public. We wanted to see how they would group the existing response options under the impairment question. We chose to do this research because the UK Statistics Authority Strategy emphasises the importance of being inclusive, so we wanted to ensure the groupings were made based on the views of the public. Findings from the card sort demonstrated a high level of agreement for either 3 high-level groupings or 2 high-level groupings. The 3 groups are as follows:

  • physical impairments — which includes “dexterity”, “mobility”, and “stamina”
  • sensory impairments — which includes “hearing” and “vision”
  • mental or cognitive impairments – which includes “memory, “mental health”, “learning or understanding or concentrating”, and “socially or behaviourally”

To create 2 high-level groups the physical and sensory impairments can be combined.

In our short-term next steps, we will publish a harmonised approach for grouping these impairments and will monitor their statistical performance with the help of our users.


The unmoderated card sort exercise was run using Optimal Workshop. It was an electronic sorting exercise. Participants were recruited using People for Research’s recruitment services with the specified criteria that participants must:

  • live in the UK
  • be over 16 years old
  • have access to a computer or mobile device

We had 1131 respondents who completed two unmoderated card sort tasks. One card sort task was for ethnicity and the second card sort task was for disability. The respondents then provided demographic information.

In the card sort, participants were shown all the cards at once in a randomised order. Each card presented one impairment from the response option list, taken from the impairment harmonised standard. Participants were instructed that the aim of the task was to help us understand how they felt different impairments relate to each other. The following instructions were provided:

“We would like you to sort these cards into groups that make sense to you. There is no right or wrong answer. Just do what comes naturally. You may feel that some cards do not fit with any others. If this is the case, you can put these in a group of their own. Please try not to do this with too many cards.”

Participants were tasked with dragging cards from the list on the left into open space in the right. The first card moved over created a group. Participants could add other cards to this group by dragging them on top of the first card and then labelling the group. Participants were presented the following impairment list in a randomised order:

  • Vision – for example blindness or partial sight
  • Hearing – for example deafness or partial hearing
  • Mobility – for example walking short distances or climbing stairs
  • Dexterity – for example lifting and carrying objects, using a keyboard
  • Learning or understanding or concentrating
  • Memory
  • Mental Health
  • Stamina or breathing or fatigue
  • Socially or behaviourally – for example associated with autism spectrum disorder (ASD) which includes Asperger’s or attention deficit hyperactivity disorder (ADHD)

Participants were also asked to use the “leave a comment” function to tell us about:

  • any cards they would expect to see
  • any cards they thought were missing
  • what group they felt any missing cards would belong to
  • whether any of the wording did not make sense and what wording they would find more helpful

After completing the card sort, participants were then asked socio-demographic questions using the relevant harmonised standards for:

  • Age
  • national identity
  • ethnicity
  • religion
  • UK country of residence
  • long lasting health condition or illness
  • activity restriction
  • impairments

Participants are considered disabled under the Equality Act 2010 definition if they answer “yes” to the long-lasting health conditions and illnesses standard and “yes a lot” or “yes a little” to question 1 of the activity restriction standards.


The main aim of the unmoderated card sort was to understand how people would group and label the impairment response options from the impairment harmonised standard.

The most common response from participants was to split the impairments into 3 groups. 31% of respondents chose to split the list into 3 groups. The groups were labelled “Mental or Cognitive Impairments”, “Physical Impairments” and “Sensory Impairments”. The groups were:

  • “Mental/Cognitive impairments” – this group included impairments in the categories of “Learning or understanding or concentrating”, “Memory”, “Mental Health”, and “Socially or Behaviourally”
  • “Physical Impairments” – this group included impairments relating to “Dexterity”, “Mobility”, and “Stamina or breathing or fatigue”
  • “Sensory Impairments” – this group included impairments relating to “Hearing” and “Vision”

30% of respondents preferred to split the list into 3 groups. The groups were labelled “Mental/Cognitive Impairments” and “Physical impairments”. The groups were:

  • “Mental/Cognitive impairments” – this group included impairments in the categories of “Learning or understanding or concentrating”, “Memory”, “Mental Health”, and “Socially or Behaviourally”
  • “Physical Impairments” – this group included impairments relating to “Dexterity”, “Hearing”, “Mobility”, “Stamina or breathing or fatigue”, and “Vision”

We also investigated whether people who meet the definition of “disabled” in the Equality Act 2010 grouped the impairments differently to non-disabled people. Our findings showed that there was no clear difference. People who were disabled equally preferred 3 groups and 2 groups. 30% of disabled participants chose to split the list into 3 groups and another 31% of disabled participants chose to split the list into 2 groups. People who were not disabled showed a slight preference for 3 groups compared to two groups. 31% of non-disabled participants chose to split the list into 3 groups and 29% of non-disabled participants chose to split the list into 2 groups.

We also found that disabled people who reported having an impairment, grouped and labelled that impairment the same way as people who did not report having that impairment.

We explored whether there was a difference between these groups for each impairment. The only notable difference was for the hearing impairment. 30% of all respondents grouped hearing under “physical impairments”, often with “dexterity”, “mobility”, “stamina or breathing or fatigue” and “vision”. 33% of respondents who were disabled grouped hearing under “physical impairments”, often with “dexterity”, “mobility”, “stamina or breathing or fatigue”, and “vision”. But people who were disabled and did report a hearing impairment were more mixed in their responses:

  • 20% of these participants labelled “hearing” as a “sensory impairment” and grouped it with “vision”
  • 18% of these participants labelled hearing as a “physical impairment” and grouped it most often with “dexterity”, “mobility”, “stamina or breathing or fatigue”, and “vision”

The same was not true for respondents who were disabled and reported having a vision impairment. 30% of all respondents, 34% of disabled respondents and 27% of disabled respondents with a vision impairment grouped it under “physical impairments” with “dexterity”, “hearing”, “mobility”, “stamina or breathing or fatigue”, and “hearing”.

To see whether our high-category labels created groups of equal size we explored impairment frequencies using the 2020 Opinions and Lifestyles Survey. There were 75,448 reported impairments. Our results showed that 53.5% of impairments were in the physical grouping, 34.2% were mental or cognitive and 12.3% were in the sensory grouping. The smaller sensory group could therefore still pose disclosure issues in analysis, in which case it can be grouped with physical impairments.

We have also considered the prevalence of co-occurring impairments that cross the high-level groupings outlined above. We explored data from a recent publication which shows linkages between impairment types. After removing cases of ‘missing’ or ‘other’ impairment responses, we found that 25.4% of cases did have multiple impairments across the 3 groups. 74.6% of cases did not. For the suggested 2 groups: 14.4% did report at least one physical or sensory impairment and one mental or cognitive impairment. 85.6% of cases reported impairments within one of the groupings only. These results highlight that our proposed high-level groupings may help to disaggregate impairment data where disclosure would otherwise prevent publication. But they should also be used with some caution.

We have proposed these high-level categories based on public understanding of how the impairment options relate to one another. We now need to understand whether these groupings meet the needs of our stakeholders.

Interviewer discomfort

We surveyed Office for National Statistics (ONS) interviewers in 2022 and discovered that questions on the topic of impairments can make interviewers feel uncomfortable. Interviewers found the following aspects of these survey questions challenging:

  • feeling like they are being intrusive by asking about mental health
  • listening to respondents talking in-depth about their experience and having to reduce their answer to a tick box
  • worrying that the sequence of questions will make the respondent think they have not been listening – especially where respondents have already expressed, they are disabled, and later questions cover the same subject
  • listening to the answers when respondents have “chronic or life limiting” conditions or illnesses

None of these factors are specific to the harmonised standards, but it is important to be aware of this interviewer feedback. It reminds us that these survey questions ask about very personal and sometimes painful experiences, and that we should consider the wellbeing of everyone involved in collecting and giving this data.

We will continue to work with interviewers in the long-term, and especially regarding how comfortable they feel asking any newly proposed harmonised disability questions.

Known strengths of the current standard

Easy to complete the questions

Most respondents said they found it easy to complete the long-lasting illnesses and activity restriction questions. Participants in our cognitive testing in 2022 described the questions as “nice and clear”, and “easy, it’s quite straightforward”. We have more evidence that the questions are easy to complete from Office for National Statistics interviewers. In 2022 we surveyed these interviewers about the long-lasting health conditions and illness harmonised standard, and both questions in the activity restriction harmonised standard. Interviewers tended to say that they felt respondents found the questions easy to complete (60% to 74% believed respondents found the questions “very easy” or “easy”), that they were confident in their answer (70% to 79% believed respondents were ‘very confident’ or ‘confident’ in answering the question) and that the question worked well (62% to 71% believed the question worked “very well” or “quite well”). These results suggest that the questions perform well and are easy to complete for most respondents. We hypothesise that this is the case for people who are familiar with the questions and are in the “core” disabled group. The questions may be more difficult for those in the “non-core” group, which includes people who:

  • receive medication or treatment
  • are neurodiverse
  • have fluctuating conditions
  • have progressive conditions

We expect that reported ease and confidence can be further improved by clarifying the difference between “a little” and “a lot” response options in the activity restriction question.

Widespread use of questions

The questions from the harmonised standards have been used in many surveys. This has made them familiar to respondents and has led to good harmonisation of data on this topic. Many respondents expressed familiarity with the questions in our cognitive testing. One respondent said, “I feel like I get asked this a lot by now, and I am quite familiar with it, so that makes it easier to answer”.

The widespread use of the questions also brings the benefit of harmonised data which makes it easier to compare information across datasets. Our team reviewed the use of the long-lasting health condition and illness and activity restriction harmonised standards in 2020 and found 14 surveys (across England, Wales, Scotland and Northern Ireland; not including the 2021 Census for England and Wales) used both the long-lasting health condition and illness question and the first activity restriction question. The team found a further 7 surveys which used elements of the standards. The comparability this provides is important to our users. This has been made clear by the number of data users who have asked for data that is both longitudinally and cross-sectionally comparable.

Two-stage process

The two disability questions adequately capture the disabled population and work well for respondents. Although there are technically four questions in the three disability harmonised standards, many users view the core questions as:

When these two questions are asked together, they form the best way of capturing information about the core disabled population in line with UK legislation. Our research has identified two main benefits of this two-stage process.

Keeping the process short

A two-stage process is short, which means it is more likely to be used than longer question blocks. Data collectors said that two questions are manageable to add to a survey to capture disability. Longer question blocks would cost more to add, which makes them less desirable.

This also aligns with our findings from our research to understand the potential uptake of new harmonised standards, where we found data collectors were reluctant to add more questions to a survey.

Reducing the burden on respondents

Cognitive testing for Census 2021 for England and Wales found that “overall, participants preferred the two-stage questions to the one-stage question… as this reduced respondent burden and allowed for a more accurate response”. The report also notes “[P]articipants felt that the one-stage question was asking two questions instead of one and that it could not distinguish between those with no health conditions and those with a long-term condition that did not impact on day-to-day activities”.

Allows expression of multiple impairments, conditions or illnesses

The impairment question is “multi-select”, which means respondents can choose as many or as few response options as they see fit. They are not forced to choose a single area of impairment. This is a strength of the current question. As a 2022 ONS publication stated “although 3,573 of identified respondents report only singular impairment types, most individuals report interactions of two or more impairment types”. There is also qualitative data supporting this strength. Recent cognitive testing found respondents who had more than one impairment said it was helpful they could select more than one impairment.

We discovered something similar with the long-lasting health conditions and illness question. Respondents expressed positivity towards the wording of the question and how it was helpful for people who did not have a single condition or illness that is easy to classify. This included respondents who had multiple health conditions and illnesses, as the question does not assume that each person has only one condition or illness. One respondent said, “it’s good that there is a plural there because it is multiple conditions”.

Many respondents also spoke about how physical and mental often overlap. One respondent said “[P]hysical disabilities, I do have some which impact my mental health as well”. Respondents also said that a benefit of the question was that people did not have to decide whether their condition or illness was mental or physical. One respondent said “I am aware that quite often it then asks me to specify by picking one but that’s complicated”.

Areas with conflicting views

Alignment to legislation and capturing all legally disabled groups

The standards do not align with legislation exactly. The standards allow data collectors to gather information about the “core disabled population” but they do not capture everyone who is legally disabled under the Equality Act (2010). For example, it is unclear whether the standards capture information about people with progressive conditions that may or may not affect their day-to-day activities.

The Equality Act (2010) states that “certain medical conditions” are always included as a disability. These include “cancer, HIV infection and multiple sclerosis”. The Disability Discrimination Act (1995) also refers to “progressive conditions” more generally and includes specific provision for muscular dystrophy. Data on groups outside of the core disabled population was being captured elsewhere in 2008, which means the original design did not intend for these people to be counted by the disability questions. You can find out more about this in the Health Statistics Quarterly article called “Update on the harmonisation of disability data collection in UK surveys”.

Groups outside the core disabled population will only be included by these standards if they answer the questions in a certain way. We are investigating a potential solution for this issue with the Office for National Statistics social survey research and design team. The team are developing a question that will capture specific health conditions. This could be used with existing harmonised standards to create data that better aligns to legislation.

Similarly, harmonised standards may also fail to capture neurodiverse people. Neurodivergence is covered by legislation if it has “a substantial adverse effect on the ability of the person concerned to carry out normal day-to-day activities”. However, cognitive interviews suggest that neurodiverse people do not consistently identify as having a long-lasting health condition or illness. One respondent demonstrated this point by saying “I don’t see autism being in this question because it’s a neurodiversity not a mental health condition”. This means the disability questions should not be used to count the neurodiverse population. We will address this in the long-term by designing a question that will gather information about the neurodiverse population and allow data users to separate the data to study specific groups.

Some people also thought sensory impairments should not be categorised as a physical or mental health condition or illness. But our research found that people who expressed this view still knew to record themselves as the question intended. One respondent said, “blindness is not necessarily a physical or mental health condition or illness, but I knew what it was asking, it tends to be sensory, a sensory impairment”.

The harmonised standards may include groups who do not fall under the legal disability definition. Both Section 6 of the Equality Act (2010) and Section 1 of the Disability Discrimination Act (1995) refer to presence of condition or illness and how long there has been an effect on daily life. But the harmonised standards capture data on how long a person has had a condition or illness and the presence of an effect on daily life. This means that, unlike the Equality Act (2010) definition, the harmonised standards would capture information about a person with a long-term condition that only affected them for a short period. We will update our guidance in the short-term to clarify how the harmonised standards do not align to legislation.

Alignment to legislation and the effect of treatment or medication

There is also a misalignment between harmonised standards and legislation when considering the effects of treatment or medication on activity restriction. The Equality Act (2010) states “an impairment is to be treated as having a substantial adverse effect on the ability of the person concerned to carry out normal day-to-day activities if measures are being taken to treat or correct it, and but for that, it would be likely to have that effect”. The Act aims to include people who have a condition regardless of whether they are receiving “any treatment or medication”. By contrast, the harmonised standard emphasises current activity restriction and is designed to exclude cases where medication or treatment means that day-to-day activities are not affected.

The harmonised standard asks respondents whether their condition or illness “currently” affects their day-to-day activities. The guidance states “you should consider whether you are affected while receiving any treatment or medication”. A 2008 workshop showed that questions in line with legislation were confusing for respondents. It found that respondents did not like being asked to “hypothesise or speculate about their likely restriction if medication was withdrawn”. The workshop notes state that this “added unnecessary complexity and would introduce a high level of subjectivity between respondent perceptions, potentially leading to spurious results”. You can find out more about the results of the 2008 workshops in the Health Statistics Quarterly article called “Update on the harmonisation of disability data collection in UK surveys”.

Our 2022 survey of Office for National Statistics interviewers also suggests that the current guidance is not as good as it could be. We found that some interviewers were including people even when their medication meant their activity was not restricted. One interviewer said “sometimes respondents get confused and say they used to have a health problem but now they take medication so no longer have a health problem, for example blood pressure. When of course they do still have a health problem, but it is masked by the medication”. It is possible we could solve this issue by providing clearer guidance. The current guidance just says that respondents should “consider” their medication. We could take a more direct approach by reminding respondents that the question is asking about current activity restrictions regardless of whether they are receiving treatment. We aim to explore this possibility in the long-term.

Aligning with other models of disability

The three most applied models of disability are the:

  • medical model
  • social model
  • biopsychosocial model

The medical model concentrates on symptoms, conditions, or illnesses. The model has been criticised for attributing the disability to the health condition or illness that a person has.

The social model states that people with impairments are not disabled by the impairment itself, but rather by societal barriers. The social model concentrates on what a person needs to overcome these barriers, not how to fix the impairment. Scottish Government and Welsh Government have both named the social model as their preferred approach. Many disabled people have also expressed their preference for the social model.

The biopsychosocial model views disability as the result of health, personal and environmental factors. It is the model supported by the World Health Organisation and recommended by the Inclusive Data Taskforce (IDTF).

We need to consider how well the harmonised standards align to these models based on stakeholder feedback and the recent IDTF recommendations. While some aspects of the standards fit well with the social model, others do not.

Historically, the long-lasting health conditions and illness harmonised standard asked if people “have” a disability. This suggests a person’s condition is what makes them disabled, rather than the societal barriers which can prevent access to certain things. The language has been updated to ask whether people “have any physical or mental health conditions or illnesses”. But these terms are criticised as being too medical, compared to the preferred term “impairment”.

The activity restriction harmonised standard cannot be reconciled with the social model of disability. This is because the question refers to disability as a result of a health condition or illness a person may have, rather than as a result of societal barriers.

The impairment harmonised standard aligns with the social model of disability to some degree. The question concentrates on impairments rather than health conditions. But the guidance for the “socially or behaviourally” response option does mention specific medical diagnoses. Some users have expressed interest in data on the societal adjustments needed, rather than what domains of a person’s life are affected by their impairment, or impairments.

The current harmonised standards have been designed to align to UK legislation. This means they do not fully account for the social or biopsychosocial models of disability. We would need to add many additional questions to capture the biopsychosocial model in its entirety, and our research has shown that stakeholders are not keen to add more questions. When we asked stakeholders about the need to align to the biopsychosocial model, they often said they preferred the social or medical models. They did not think it would be possible to incorporate all the different elements of the biopsychosocial model into a single standard.

It is also important to note that UK legislation and the social model of disability are incompatible in some ways. For example, the Equality Act (2010) considers certain people, such as people with progressive conditions or receiving treatment, to be disabled regardless of whether they encounter barriers. The harmonised standards will continue to align to legislation, but we will also try to incorporate elements of these models wherever possible. In the long-term we will consider an additional question that is aligned more closely to the social model of disability. This work will depend on the findings from ongoing research by the Welsh Government’s Disability Disparity Evidence Unit who are currently assessing the feasibility of capturing data according to the social model.

The term “disabled”

The term “disabled” is contentious. People who endorse the social model are pleased that we have stopped using the term “disability” in favour of “long-lasting health conditions and illnesses” and consider this as progress. Users have expressed a need to avoid combining the concept of having an impairment with being disabled. Not everyone who is covered by the Equality Act (2010) definition of disability identifies with the term “disability”. Testing during development found that the term “disability” was viewed as “a technical or official label linked to benefits such as Disability Living Allowance and the “Blue Badge”, with some respondents uncertain about applying it to themselves in the absence of benefits of some kind”. Data users representing community groups expressed a similar point of view during recent research for Census 2021 England and Wales. The research found “some considered that the “disability” term might be less inclusive to those with milder long-term health conditions and illnesses. “Disability” was also considered to be more linked to physical than mental health conditions”.

There are user groups who prefer the terms “disabled” or “disability”. Third sector groups, including Disabled People’s Organisations, expressed their preference for these terms in the research. They said that these terms are important because of those who self-identify as disabled. Some research participants expressed a concern that people who are considered disabled under the Equality Act (2010) may not identify as having long lasting health conditions and illnesses. We will continue to explore the term “disabled” and its implications that as part of our work in the medium-term.

Continued use of the current standard

We would encourage users to continue to use this standard while we work on it, despite the current limitations. This is because it was developed using research and testing that was conducted for Census 2021 in England and Wales. This research and testing process was thorough, which means it is likely this question will still outperform others that did not undergo such a development process.


Our initial research and engagement summarised in this review found that the current harmonised standards relating to disability have many strengths but also a range of weaknesses. We have worked with a wide range of users across the private, public and third sector as well as using existing evidence and undertaking our own research. We have identified a broad range of needs, but many of them are contradictory. This means they cannot all be met by a single harmonised standard.

These findings were used to inform the current review and will be used to update the harmonised standards accordingly. We cannot address all the user needs discussed, so we have prioritised and addressed them in our review.

Wording of questions

Our research found that data providers were concerned about the wording of the questions. Firstly, they highlighted the need for the disability question to be easier to understand and to have clearer boundaries to reduce subjectivity. For example, providers want to know whether they should include self-diagnosis or just formal diagnosis.

Secondly, data providers who are disabled expressed a need for more inclusive, consistent, and familiar language in the long-lasting health question. Some data providers felt like it was hard to self-identify without the term “disabled” in the question.

Finally, both data collectors and data providers said that the questions should be unintrusive. They explained that if the questions made respondents uncomfortable this could cause non-responses.

Diversity of disability

Data users and collectors both indicated a need for the long-lasting and day-to-day activity questions to capture information about the diversity of disability. For example, there is a need to capture people who do not identify as disabled but would say that their daily activities are substantially affected. There is also a need to capture disability that does not fit with the requirement to have been affected for “at least 12 months”, but still classifies as disabled. For example, this could refer to a disability that does not meet the definition in the Equality Act.

Government analysts need impairment data on currently under-represented groups, such as people who have both a physical and mental condition, or a condition that does not fit in either of these categories. These analysts want to ensure that they are being inclusive in their findings. They also asked for data that considers more than one condition to understand how different disabilities overlap. This is also supported by stakeholder needs from data providers with multiple and complex conditions. These data providers said they would prefer to be able to express their physical and mental conditions or illnesses in a single response. This would allow people to express their multifaceted experiences and help to avoid splitting conditions into either physical or mental.

Some data collectors, users and processors mentioned a need for already existing detail, such as counts split by disabled and non-disabled, severity and different types of disabilities. But others expressed a need for unreported information. These requests varied by type of user and organisation depending on their work requirements. There were many requests for a question that looks at or considers:

  • health conditions as well as impairments
  • extra detail in the impairment question – including adding more impairments and more detail around the existing categories
  • visible and non-visible conditions
  • conditions that change over time
  • a new category option to indicate whether someone has had the impairment their whole life

Service providers also wanted to be able to capture disability related to ageing and to explore the wider effects of disability, for both older and younger people.

Guidance and examples

Many data users and providers expressed a need for better guidance and examples. Many users said the impairment categories can be misleading, incorrect, and inconsistent. For example, certain impairments provide examples to help the respondent answer more appropriately, but others do not provide any additional information. This makes it harder for the respondent to identify with these impairments. Lots of users have asked for guidance to be included for all impairments.

We also discovered a need for the activity restriction question to define the response options for “a little” and “a lot” because the categories are currently subjective. Data users from charity and non-government organisations also found the impairment categories to be outdated and inappropriate. These users expressed a need for them all to be revisited and researched appropriately.


We identified a user need to keep the disability standard in line with the appropriate legislation. Data analysts expressed a need to be able to identify disability in line with a definition used in UK policy.

Different data users also expressed an interest in having a definition that could produce data that is comparable with the Scottish Census. For example, data analysts and charity organisations expressed that the Scottish impairment list is much more detailed than the current impairment standard. Individuals are also able to distinguish different groups within the high-level groupings.

The medical model and the social model of disability

Research participants also discussed the need for the disability standard to align with the medical and social models of disability. They saw that the current standard largely follows the medical model and a few were happy with this alignment. But most data collectors, providers and users identified a need for the definition to align more closely with the social model. These participants said that disability and impairment should be seen as separate concepts. Not all people with disabilities classify themselves as being impaired and not all people with impairments identify as disabled. Data users also said that the question should move blame away from the condition or illness and instead concentrate on how society has impaired the individual by creating barriers that exclude people with impairments or disabilities. This would help align the standard with the social model. The medical approach was seen to be outdated and did not consider external factors.

Stakeholders were asked explicitly about the IDTF recommendation to follow the biopsychosocial model. Data providers, users and collectors reported that they were not familiar with the concept and asked for more information about it. Most stakeholders preferred the social or medical model after the biopsychosocial model was explained to them. This was because they did not understand how it would be possible to incorporate all the different elements from the model into a single standard.

Both government and non-government researchers expressed a need for data on access needs and requirements so that they could understand the needs of different types of disability. They have suggested that the activity restriction question should ask which activities are restricted rather than just how much. By doing this they will be able to target specific services to improve quality of life for disabled people.

Neurodivergent conditions

Policy professionals and researchers expressed the need for the disability standard to consider neurodivergent conditions. Users need a better understanding of different forms of neurodiversity . For example, they need to examine data by protected characteristics and across neurodivergent conditions, including the need for people to report more than one condition or learning disability. This is important as currently someone with a neurodiverse condition could answer “yes” to all the impairment questions.

Data coherence

Policy makers, government officials and data analysts expressed the need to have consistency of data. This would enable them to analyse data trends over time. They also need data that is comparable across datasets and the devolved nations.

Government analysts and policy advisors also asked for longitudinal data that tracks individuals through their life so that we can:

  • have a better understanding of what it means to be disabled
  • explore the different transitions in and out of disability

This is not something we can capture through a single question. But we can include changes that may help us capture this data better, such as including a question that asks when the disability began or stopped. A few stakeholders said that any changes to the question would limit coherence of data. But others said that having a small change, changes with a warning, or a new question alongside the original would help to support this.

Survey of data collectors

Our survey of data collectors and producers revealed some important technical and logistical issues with using and updating harmonised standards.

It is not easy to implement changes, but 47% of participants are open to change under the right circumstances. When proposing changes we should remember that:

  • stakeholders must consulted and we must be able to demonstrate strong operational needs for any changes — we expect it would take 12 months to agree changes and 12 months to implement them
  • changes will be restricted by maintaining time series and length of the survey — changes are often limited to certain time periods depending on the life cycle of the collection
  • additional data collection is not necessarily an issue because most users use analytical programming software to a good degree, or it is at least possible to use software for collection – it is important to remember that the programming of small surveys is limited by the tools used
  • approximately 25% of data collectors and users have the capacity and expertise to process free text — it is possible to collect free text, but rarely possible to process it due to resource limitations
  • we must consider the collection mode and subsequent effects of it, including any effects on user experience — changes must be feasible and tested that Data Protection, GDPR and Equality Act cover most of the legislation requirements. administrative data systems can be complicated — the systems are not always managed by the data producers and often include multiple layers of stakeholders

We will use this information to inform any updates to the harmonised standards.

What happens next

We have reviewed our research and engagement work to create plans for the short, medium, and longer-term.

In the short term (Spring 2023), we will:

  • publish a warning with the activity restriction standard explaining the potential issues with disaggregating “a little” from “a lot”
  • research further into whether “continence” and “communication” should be included in the impairments questions
  • publish a harmonised approach to grouping impairments and monitor the use and performance of these new high-level categories in the long-term
  • clarify our published guidance on measuring disability for the Equality Act (2010) — we aim to make it clear to users in what ways the standards do not align to legislation, please note that some of this is already on the webpage and also applies to the Disability Discrimination Act (1995) for Northern Ireland
  • clarify how the harmonised standards align to the medical and social models of disability on our webpages

In the medium term (2023), we will:

  • change and test the guidance on the impairment response option for “socially or behaviourally”
  • design and test guidance for the impairment response options that do not have guidance already
  • design and test improvements for the impairment examples that have been highlighted as confusing
  • explore the use of the term “disabled” and the implications it may have
  • work with disabled people to better understand their experience, and work with people who are not disabled to better understand how they answer existing questions
  • publish updated questions for online self-completion

In the longer term (2024), we will:

  • review and optimise our questions according to delivery mode, concentrating on telephone and face-to-face mode questions
  • research the possibility of designing guidance to define and separate the activity restriction options “a little” and “a lot”
  • continue to work with survey interviewers and ask for their feedback specifically on how comfortable or uncomfortable they would feel asking any new questions
  • design new questions to meet identified gaps, such as a question to capture information about the neurodiverse population and separate the data to enable users to study specific aspects of neurodivergence
  • explore how we can work with the Office for National Statistics Social Survey Transformation division to learn from their research on capturing specific health conditions. This could be used to inform how we capture the non-core disabled population in line with UK legislation
  • design and test improvements to the guidance for the activity restriction question to make it clear activity restriction should be considered regardless of current treatment or medication
  • consider the creation of a new question aligned to the social model – this will depend on ongoing work from the Welsh Government’s Disability Disparity Evidence Unit
  • work with stakeholders to develop harmonisation guidance for administrative data sources

Contact the Government Statistical Service (GSS) Harmonisation Team

If you would like more information about the disability harmonised standard, please contact the team at

You can find more information about Harmonisation on our webpage.

  • If you would like us to get in touch with you then please leave your contact details or email directly.
  • This field is for validation purposes and should be left unchanged.