Primary Care Dementia Data: Consultation 2026 (now closed)

The consultation (details below) is now closed. Our detailed response can be found in Primary Care Dementia consultation 2026 - Response.

The publication in it's updated format will be published on our Primary Care Dementia Data series page from Thursday 30 July 2026.

If you have any questions about the consultation contact us at: [email protected].

The Primary Care Dementia Data consultation ran from Thursday 19 March to Wednesday 15 April 2026 and was guided by the need to align available resources with user priorities. The changes proposed aimed to streamline the publication process and outputs to ensure operational optimisation and efficiency whilst ensuring that user priorities are accounted for, and that NHS England still meets statutory obligations regarding publication of data.

NHS England engaged with key stakeholders ahead of this consultation to determine the fundamental requirements of the PCDD publication and aimed to incorporate this feedback into the proposals. The consultation does not impact the status of the publication as an Official Statistic.

This consultation proposed:

• a reduction in publication frequency
• a reduction in publication scope and complexity

NHS England and the Department for Health and Social Care are working with stakeholders to identify a suitable alternative route to provide further analyses of dementia data as part of a wider re-design of the way that intelligence on dementia will be produced and disseminated.

It was proposed that from 2026/27, the PCDD publication will be published on a quarterly basis (April/July/October/January) rather than a monthly basis.

This proposal was driven by a change in the collection frequency of the Core GP Contract (CGPC) GPES extract, which is the underlying data source for this publication. The CGPC extract was previously collected on a monthly basis however from April 2026, this collection became quarterly. To align with the latest CGPC collection, the PCDD publication will be moved to quarterly rather than monthly. This will not reduce data availability, it purely reflects the change in the underlying data source.

The data are relatively consistent from month to month therefore a reduction in frequency and therefore less timely data was not anticipated to impact the utility of the data itself.

Due to a change in the collection frequency of the underlying dataset, this proposal is a must and cannot be avoided.

Following the consultation and a review of the feedback, this proposal has been upheld. The PCDD publication will be quarterly from 2026/27. Please see the consultation response for further information.

It was proposed that only data for a single quarter will be published with each release of the PCDD publication, as opposed to the 13 month time series previously published with every release.

Since historical data remain unchanged in each release, reissuing the previous 12 months simply duplicates existing information. By publishing only a single quarter’s data, unnecessary repetition is effectively minimised.

Please note that users relying on the time series insights are still be able to generate them using the published data.

Following the consultation and a review of the feedback, this proposal has been upheld. From 2026/27, only data for a single quarter will be published with each release. Please see the consultation response for further information.

It was proposed that NHS England will no longer publish the national excel summary file which contained the following data:

1. National level counts (including a 12 month time series) of practice/patient coverage, recorded dementia diagnoses, dementia diagnosis rate estimates, assessment and referral counts, medication and care plan review counts, antipsychotic prescribing counts, incidence and young onset counts, delirium, comorbidity, palliative care and cognitive impairment counts.
2. National level age and sex breakdowns of patients with a recorded dementia diagnosis
3. National level ethnicity breakdowns of patients with a recorded dementia diagnosis
4. National level breakdowns of dementia type, for patients with a recorded dementia diagnosis
5. National level breakdowns of residence type, for patients with a recorded dementia diagnosis
6. Dementia diagnosis rates at national, regional, ICB, sub ICB and Local Authority levels
7. Antipsychotic prescribing counts at national, regional, ICB and sub ICB levels

The underlying data will still be published in an alternative format and all constituent data will continue to be published.

Following the consultation and a review of the feedback, this proposal has been upheld. The national excel summary file will no longer be produced from 2026/27. Please see the consultation response for further information.

It was proposed to cease production of the Dementia Diagnosis Rate (DDR) indicator at local authority (LA) level, and to only produce this for the following NHS geographies:

• Sub ICB
• ICB
• Region
• National

The DDR was originally calculated for both NHS and LA geographies, as dementia care is commissioned across NHS and local government depending on care pathways and commissioning structures. The DDR calculation is however computationally resource intensive and requires producing for each level of geography individually, that is, one iteration per geography level such as Sub ICB, or LA.

As the data are collected at practice level, current mappings to local authority are carried out by mapping the postcode of the GP practice to manually downloaded local authority reference data produces by the Office for National Statistics (ONS); practice level data are then aggregated accordingly. Due to the limitations of postcode mapping, this approach has low clarity for the data users and may not necessarily represent the local authority or Government Office Region (GOR) under which the patient lives, or dementia care is commissioned.  

Alternatively, mappings to NHS geographies are direct relationships between GP practices and the Region, ICB or Sub ICB whom they are commissioned by. These mappings are maintained within corporate reference data by the Organisation Data Service (ODS) and the Reference Data team within NHS England, and are fully automated.

The proposal to only calculate and publish the DDR at NHS geography levels and to cease production of the DDR for LA geography levels results in efficiency savings whilst maintaining a direct commissioning line and the ability to use the DDR for planning and research across different NHS commissioning geographies.

Following the consultation and a review of the feedback, this proposal has been withdrawn. The DDR will continue to be calculated and published for both NHS and LA geographies. Please see the consultation response for further information.

It was proposed to publish all data at sub ICB level only and to remove all disclosure control from the publication outputs with no rounding or suppression applied.

Due to the addition of various measures at different points in time, there were several methodologies for disclosure control within the PCDD publication outputs which include differing rules for rounding and suppression depending on the particular measure and geography level at which the data are published. This proposal aimed to align disclosure control methodologies whilst still protecting confidentiality.

The disclosure control methodologies in place at the time of this consultation are described in table 1 below:

The removal of suppression and rounding of this data once it is aggregated to sub ICB level allows for a consistent approach to disclosure control across the entire publication whilst ensuring that data which was previously heavily suppressed can actually be used for additional purposes, and increasing the accuracy of any current research/analysis which makes use of these measures.

Disclosure control exists for the purpose of protecting confidentiality therefore the following considerations have been taken into account with regards to this proposal:

1. The aggregation of data to sub ICB level increases the underlying population to >100k for any given sub ICB therefore the risk of small numbers is reduced. Where small numbers do occur the main risk of disclosure control is then due to self-identification; this is deemed to be low risk and acceptable within these circumstances as it is not possible to determine any additional information, other than what you already know about yourself or someone else.
2. No data are published at practice level, and all measures are distinct counts relating to a specific diagnosis or care intervention, therefore it is not possible to compare or combine data across different measures to reveal additional disclosive information.
3. No personal identifiable data (PID) such as name, DOB, address is collected or published.

In addition, a full risk assessment is undertaken each year for this publication, or more frequently if changes are made.

[1] A more comprehensive description of the disclosure control methodology is available.

Following the consultation and a review of the feedback, this proposal has been amended. With the exception of antipsychotic prescribing measures, any available measures which are currently published at practice level will continue to be published at practice level.  Please see the consultation response for further information.

It was proposed to reduce the number of data CSV outputs from 12 CSV files to 2. There are 2 additional CSV files which provide mapping and data dictionary information which will be retained. A comparison of the proposed files can be found in tables 2 and 3 below along with the rationale for each proposed change.

The consolidation of several CSV files into a single CSV file output provides the following benefits:

• Standardised output format which is more consistent with other publications of GPES extracts such as the GP Contract Services, and Network Contract Direct Enhanced Services (NCDES).
• Reduced duplication of ‘contextual’ measures across output files, for example, the dementia register, or patient list size are currently repeated across several CSV files for use as a denominator or comparator for other measures.
• Where users want to look at several measures there is no need to access several files as all data will be within a single file and users can select only those measures they are interested in.

Following the consultation and a review of the feedback, this proposal has been amended. Output files will still be consolidated, however there will be now be one sub ICB level file, one practice level file, and 2 supporting information files (mapping and data dictionary).  Please see the consultation response for further information.