Welsh Index of Multiple Deprivation (WIMD) 2025 technical report - Health domain

The purpose of this domain is to measure lack of good health in all people, and to capture predictors of future health based on deprivation experienced in childhood. Some indicators in this domain are age-sex standardised to account for population differences across small areas. The domain has a relative weight of 15% in the overall index.

Limiting long term illness

Type of indicator

Rate per 100, indirectly standardised for the age and sex profile of the population.

Numerator

Number of people with any long-term illness, health problem or disability that limits daily activities or work.

Denominator

All usual residents.

Source and time period

Census 2021, Office for National Statistics (ONS).

Comparability with WIMD 2019

Comparable.

Premature death rate

Type of indicator

Rate per 100,000, indirectly standardised for the age and sex profile of the population.

Numerator

Number of deaths of those under the age of 75.

Low birth weight

Type of indicator

Percentage.

Numerator

Number of live single births less than 2.5 kg (5.5lb), which is classified as a low birth weight.

Denominator

Number of singleton live births.

Source and time period

ONS birth registrations, 2015 to 2024.

Comparability with WIMD 2019

Comparable.

Denominator

All people under the age of 75.

Source and time period

Numerator: ONS death registrations, 2015 to 2024.

Denominator: Small Area Population Estimates (SAPE), ONS, mid-2013 to 2022

Additional notes

Poor health can manifest itself in lower life expectancy, which can be captured through age and sex standardised death rates.

Comparability with WIMD 2019

Comparable.

Chronic conditions diagnosed by GPs

Type of indicator

Rate per 100, indirectly standardised for the age and sex profile of the population.

Numerator

Number of people with a diagnosis for conditions from a defined list of chronic health conditions (see the additional notes for details).

Denominator

All people.

Source and time period

Numerator: Digital Health and Care Wales (DHCW), 1 July 2025.

Denominator: SAPE mid-2022, ONS; minus prisoner numbers, Ministry of Justice via ONS. 

Additional notes

The numerator was based on counts of people with diagnoses for conditions from a defined list of disease registers and sub-indicators obtained from GP practices in Wales. It measures the number of people with a current diagnosis of one or more of the conditions listed below:

• Coronary Heart Disease
• Chronic Obstructive Pulmonary Disease
• Stroke and Transient Ischaemic Attack
• Peripheral Arterial Disease
• Chronic Kidney Disease
• Diabetes Mellitus (type 1 for all ages, type 2 and other types for people aged 17 and above)
• Epilepsy

These counts were de-duplicated so that patients with more than one condition were not counted twice. Patient level data was aggregated to small areas, according to patient addresses, so that prevalence is based on where people live rather than where they are registered with a GP. 

See annex 4.1 for more detail on this indicator. 

Comparability with WIMD 2019

Comparable.

Mental health conditions diagnosed by GPs

Type of indicator

Rate per 100, indirectly standardised for the age and sex profile of the population.

Numerator

Number of people with a diagnosis for conditions from a defined list of mental health conditions (see the additional notes for details).

Denominator

All people.

Source and time period

Numerator: Digital Health and Care Wales (DHCW), July 2025.

Denominator: SAPE mid-2022, ONS; minus prisoner numbers, Ministry of Justice via ONS. 

Additional notes

The numerator was based on counts of people with diagnoses from a defined list of disease registers and sub-indicators obtained from GP practices in Wales. It measures the number of people with a current diagnosis of one or more of the conditions listed below:

• Depression
• Low mood (patients with record of low mood and an active repeat prescription for an anti-depressant)
• Anxiety disorder (including panic disorders)
• Dementia
• Severe mental illnesses (schizophrenia, bipolar affective disorder and other psychoses)

These counts were de-duplicated so that patients with more than one condition were not counted twice. Patient level data was aggregated to small areas (LSOAs), according to patient addresses, so that prevalence is based on where people live rather than where they are registered with a GP.

See annex 4.1 for more detail on this indicator.

Comparability with WIMD 2019

Comparable.

Cancer incidence

Type of indicator

Rate per 100,000, indirectly standardised for the age and sex profile of the population.

Numerator

Count of all cases of cancer includes all malignancies, excluding non-melanoma skin cancer.

Denominator

All people.

Source and time period

Numerator: Welsh Cancer Intelligence & Surveillance Unit (WCISU), PHW, 2012 to 2021.

Denominator: SAPE, mid-2012 to 2021, ONS.

Comparability with WIMD 2019

Comparable.

Low birth weight

Type of indicator

Percentage.

Numerator

Number of live single births less than 2.5 kg (5.5lb), which is classified as a low birth weight.

Denominator

Number of singleton live births.

Source and time period

ONS birth registrations, 2015 to 2024.

Comparability with WIMD 2019

Comparable.

Children aged 4 to 5 living with obesity

Type of indicator

Percentage.

Numerator

Number of children attending reception class aged 4 to 5 living with obesity.

Denominator

Total number of children attending reception class aged 4 to 5.

Source and time period

Child Measurement Programme (CMP), Public Health Wales (PHW), 2016/17 to 2018/19, combined with 2022/23 to 2023/24 (5 years in total, but see below for explanation of the gap in the middle).

Additional notes

Pupil’s home addresses were used to identify the LSOA in which children live rather than the LSOA of their school. Data for previous years also included in the WIMD 2019 indicator (2016/17 and 2017/18) has been retrospectively updated to allocate 2021 LSOAs to the records.

Obesity is calculated using the age and sex-specific body mass index (BMI) centiles (which includes height information) calculated using the British 1990 growth reference (UK90) (from a method proposed by Cole et al (1995)). Children living with obesity are those who fall in the 95th centile or above.

Due to COVID-19 pandemic impacts on data collection the CMP datasets were not available across all of Wales between 2019/20 and 2021/22. Five years of data were combined to provide robust LSOA level estimates, consisting of three years pre-pandemic and two years post-pandemic data.

The smallest level at which PHW have published CMP data since the pandemic is Primary Care Cluster level. This is because of concerns around the possible misuse of data to identify specific areas where the highest percentage of children living with obesity live and the possibility of identifying individuals when drilling down to small numbers. For these reasons, whilst LSOA level rates are used in the domain and index calculations, we do not publish them as part of WIMD indicator datasets.

Users can access data for MSOAs from PHW with the latest available for 2014/15 to 2018/19 (due to the impact of the pandemic and requirement for 5 years of data). Data for larger geographies including primary care clusters, local authorities and local health boards are available on an annual basis in CMP reports and data (PHW).

Comparability with WIMD 2019

Broadly comparable.

There are 7 indicators in the health domain, weighted as follows. Factor analysis was used to calculate the indicator weights.

• 32% Limiting long term illness
• 28% Premature deaths
• 20% Chronic conditions diagnosed by GPs
• 8% Mental health conditions diagnosed by GPs
• 5% Cancer incidence
• 4% Children aged 4 to 5 with obesity
• 3% Low birth weight

The domain has a relative weight of 15% in the overall index.

Indirect standardisation for the age and sex profile of the population was applied to five of the indicators, to adjust for different age and sex distributions between small areas. For example, we should expect to observe a higher rate of deaths in an area with an older population compared to an area of mainly young families. Standardisation attempts to adjust for these differences in population (see annex 4.2).

For every indicator, each area was ranked in order, with the most deprived area ranked 1 and the least deprived area ranked 1,917. These ranks were assigned to a normal distribution, with low ranks receiving a low normalised value. Factor analysis was then used to calculate the indicator weights. As with all domains, the final domain ranks were exponentially transformed, to form domain scores for use in the calculation of the overall WIMD 2025.

There have been some small methodological changes to the health domain between WIMD 2019 and WIMD 2025.

Limiting long term illness

The wording of the Census question was updated between 2011 and 2021 to better align with the social model of disability. However an analysis of the data shows a high degree of correlation between 2011 and 2021 data suggesting this has not significantly changed the nature of the indicator.

Due to suppression of some 2021 Census data at LSOA level the age groups used as part of the indirect standardisation process for this indicator were changed slightly for WIMD 2025.

Chronic conditions and mental health conditions diagnosed by GPs

These indicators are calculated on a similar basis as before. Data for these indicators were drawn together from GP practice systems by DHCW. During 2024 and 2025 some GP practices were migrating data systems, and changes were also being made to the data extraction tool. Although this may have had some impact on data quality in some areas, overall the indicator data trends were found to be well correlated with those for WIMD 2019 so we have no reason to suspect a significant impact.

See annex 4.2 for more information on indirect age-sex standardisation applied to some of the health indicators.

COVID-19 pandemic

The data for each indicator has been investigated to assess the impact of the COVID-19 pandemic on trends and data quality. 

The data for one indicator, children with obesity, was interrupted by the pandemic meaning no data for 2019/20 to 2021/22 across the whole of Wales was available.

Limiting long term illness was self-reported during March 2021, just over a year since the pandemic was declared. However data was found to be well correlated with Census 2011 trends used in WIMD 2019.

Two other indicators, cancer incidence and premature deaths, have ten year rolling timespans extending beyond the early phase of the pandemic (ending in 2021 and 2024 respectively). Some longer-term effects of the COVID pandemic are likely to be present in the data (albeit combined with a longer run of pre-pandemic data) and we make no adjustments for this, apart from the standardisation for age and sex profiles described above.

Overview

Disease registers are lists of patients registered at general practices who have been formally diagnosed with a disease. Since 1 October 2023, practices in Wales have been required to maintain disease registers as part of the core General Medical Services core contract.

The types of diseases for which there are formal registers are the same as those originally specified in the Quality and Outcome Framework (QOF) in 2007, which was later replaced by the Quality Assurance and Improvement Framework (QAIF) in 2019. While not part of contractual arrangements, practices also record patients who are diagnosed with underlying conditions through ‘read codes’ and SNOMED codes.

There may be some small variations in how patients are diagnosed and recorded at practice level; however, all practices are contractually responsible for maintaining high quality registers, and therefore data quality is thought to be broadly very high.

Digital Health and Care Wales (DHCW) are responsible for hosting the ‘Audit +’ system which allows access to disease register and underlying read code/SNOMED code data. Following a formal request which was approved by DHCW’s data quality system group, Welsh Government was provided access to this data at the patient LSOAs level, for inclusion in WIMD.

Data specification

DHCW provided us with counts of patients with current diagnoses (as at 1 July 2025) for one or more selected conditions, separately for mental health and chronic health conditions. This included patients who had a diagnosis at any time period prior to July 2025, as long as they were still on the register. 

Data was received from all general practices that were active on the reference date in Wales, then aggregated to the Welsh Lower Layer Super Output Areas (LSOAs), according to patient’s home address. 

The specification for the conditions included was unchanged from WIMD 2019. It aims to include conditions that are:

• less able to be managed by controls or treatment which allow the individual to lead a normal life, and
• more likely to cause substantial pain and severe disability, and are associated with decreased life expectancy.

Mental health conditions

• Depression
• Low mood (patients with record of low mood and an active repeat prescription for an anti-depressant)
• Anxiety disorder (including panic disorders)
• Dementia
• Severe mental illnesses (schizophrenia, bipolar affective disorder and other psychoses)

Chronic health conditions

• Coronary Heart Disease
• Chronic Obstructive Pulmonary Disease
• Stroke and Transient Ischaemic Attack
• Peripheral Arterial Disease
• Chronic Kidney Disease
• Diabetes Mellitus (for 0 to 16 year olds this only included diagnosis for Type 1 diabetes, due to suspected under-reporting of other types of diabetes for children; for those aged 17 and above this includes any diagnosis for diabetes)
• Epilepsy

Our WIMD 2019 technical report provides further details on investigation of conditions, and we will review the list of disease registers and sub-indicators included in future indexes.

Data quality

The way in which certain conditions are recorded across general practices may vary. How promptly patients are removed from registers when conditions are resolved may also vary between practices, and impact on our data. Broadly, a patient will be removed from the disease register by a GP when they are determined not to have the conditions anymore. However, many of the conditions in both the chronic and mental health conditions indicators are long-term and are unlikely to fully resolve.

Given the possible variation in recording practices, we compared trends against the WIMD 2019 indicator data, analysis by local authority and local health board. This revealed unsurprising patterns, with high correlation between the rates of diagnosis.

Some areas have high absolute rates of diagnosed chronic or mental health conditions, and care should be taken in interpreting this data. However, for the WIMD domain and overall index ranks, it is only the relative rank of areas on the indicator that matters. Since comparison of the ranks with previous data showed high correlation, as expected, no immediate areas of concern were identified.

General practices in Wales are in the process of periodically moving to a single software supplier for the system which records patient diagnosis data. This meant that a small number of practices were actively switching systems when the data for WIMD was extracted. This may have impacted slightly on data quality for these practices; however, when quality assurance processes were performed on the data, indicator data trends were found to be well correlated with those for WIMD 2019 and therefore we have no reason to suspect a significant impact.

Denominator (population) data

The indicators within the health domain of WIMD are indirectly age-sex standardised to adjust for the expected prevalence of disease within the underlying population. This allows the index to identify areas where health deprivations exists beyond the effect of age and sex.

For the denominator, we have used the latest available Small Area Population Estimates (mid-2022) downloaded at the time of processing (October 2025), minus the prison population (2022). This allows for the breakdowns required to standardise rates for the effect of different age and sex profiles in different areas. It is also broadly consistent with denominators used in the income, employment and community safety deprivation domains of WIMD.

Another option might have been to use the patient register data, however the data available to use would not allow us to undertake age-sex standardisation using this source. Also, the ONS have published an assessment of the quality of the NHS patient register data. This explains that the patient register has a number of issues when used for statistical purposes. The source has a number of both under- and over-coverage issues, limited audit and potential for distortive effects because of its role in GP finance. The effect of these issues will vary by geography, age and sex.

Data adjustments

There are some Welsh residents with diagnosed health conditions who would not be captured in the data recorded by Welsh GPs, and we have adjusted for these in two ways.

• There are over 13,000 Welsh residents registered with primary care providers in England. We have adjusted the rates for the 23 small areas with over a 100 residents registered in England, which account for over 12,000 of the 13,000 people affected. The standardised rates were adjusted by scaling them up to reflect the volume of residents for whom we are missing data, since we do not have information on any diagnoses made across the border.
• Prisoners are likely to remain registered with a GP at their home address rather than a GP local to their prison for the duration of their sentence. Therefore we have subtracted prisoner numbers from the population estimates (denominator) before our calculation of standardised rates.

Indirect standardisation involves applying age-sex specific rates observed at national level to the population structure of each LSOA. The reason for using age-sex standardisation for the WIMD health indicators is to adjust the indicators to allow for different age and sex distributions amongst LSOA populations. For example, one might expect to observe a higher rate of deaths in an aging population than in one consisting predominantly of young families. Standardisation attempts to adjust for these differences in population. 

The number of expected incidences (for WIMD these are limiting long-term illness, cancer, chronic and mental health conditions, and premature death) for each age-sex group in an LSOA is estimated by multiplying the number of people in the given age-sex group in the LSOA by the age-sex specific rate observed for Wales as a whole for that age-sex group. The total number of expected incidences for the LSOA is calculated by totalling the number of expected incidences for each age-sex group. The standardised ratio (e.g. of cancer incidence) for each LSOA is the number of observed incidences in the LSOA divided by the number of expected incidences. 

Standardization ratio = observed incidence / expected incidence

An indirectly age-sex standardised rate can be obtained by multiplying the standardised ratio for the LSOA by the crude rate for all of Wales. The Welsh crude rate is the number of incidences observed in Wales divided by the total Welsh population. The result is expressed as a rate per 100,000 people. 

Indirectly standardised rate = standardised ratio x Welsh crude rate x 100,000