Mental Health Services Monthly Statistics, Performance July 2025

As part of the implementation of the metrics relating to length of stay which were first published in the January 2025 performance publication, a further review of length of stay metrics was undertaken. A minor change to the methodology was implemented to select distinct spells for all metrics that present length of stay. This is to align all of the metrics relating to length of stay. 

The impact of this change is minimal and doesn't significantly impact any counts. The methodology will be included from this point forwards and will also be included in the 2024-25 end of year re-runs.

From the Provisional April 2023 data onwards, all metrics in the publication are being published in both the performance and provisional files. This is to support a move towards a single submission window towards the end of 2023.

The single submission window would mean that instead of reporting both performance and provisional data, just one cut of data would be taken. The submission window would be increased meaning that providers have longer to submit the data but that the data published would effectively be the provisional data.

As part of the July 2025 data release, a number of new metrics and breakdowns have been added to the publication. These broadly include:

• Additional Mental Health Act reporting. This reporting includes additional metrics and breakdowns. As part of the MHA annual publication (released on 18 September 2025), an enhanced dashboard will be released which will include monthly data.
• Reporting using the LD and Autism status fields. Reporting uses these fields for 4 key metrics (People in contact with services, Contacts, New Referrals and people with an open hospital spell), OAPs and MHA reporting. Caution should be taken when interpreting this data initially given the field was only introduced as part of version 6 of the dataset in April 2024 and has not previously been used until now.
• Additional metrics relating to CYP Access. This includes the addition of crude and standardised rates and comparisons between groups for ethnicity and deprivation to support health inequalities reporting at Region and ICB level.
• Minor amendments to some of the VODIM metrics. These changes have been made to better reflect guidance and reporting practices. Only Experimental DQMI scores maybe impacted by this change. 

For reporting pre April-24, the average minutes of restraint are calculated for restraints that started and ended in the month. As a result, for some restraint types, specifically seclusion and segregation, this can lead to averages that do not represent all uses of the restraint types. This is because these restraint types can occur over long periods of time, and as such cross months.

From April 2024 data onwards, these metrics will be amended to calculate the average minutes of restraint for all restraints that ended in the month. Additionally, 6 new restrictive intervention measures (MHS98a, MHS138a, MHS146, MHS146a, MHS147 and MHS147a) will be included that show the average number days (as median and mean) for incidents of seclusion and segregation that remain open at the end of the reporting period.

Full details of the methodological changes made for this publication can be found here. Changes to the code used to produce this publication are also reflected on GitHub and in the associated Metadata file.

Following the implementation of version 6 of MHSDS, some small changes have been made to the data quality metrics presented in the VODIM file within this publication:

• The construction of DQM09 has changed to reflect the fact that the Ward Site ID is now captured in the MHS903 Ward Details table. The calculation remains unchanged but the data is joined to the Ward Stay table from the MHS903 table.
• DQM45 has been removed following the removal of the Provisional diagnosis table from the dataset. 
• DQM51 has been adjusted to use data from either MHS101 or MHS102 as this data is now captured in both tables. Previously only MHS102 was used in this metric.

As part of the July 2025 release, further updates were made to the VODIM metrics, however these changes only impact experimental DQMI scores.

Referral Spell methodology

As part of the September 2024 Performance publication in the Mental Health Services Monthly Statistics publication series, a new set of metrics will be introduced which utilise a new methodology which groups sets of individual referrals into a referral spell. Using this methodology, a spell is defined as starting at the point the first referral that is in scope is received by a provider. From this point, other referrals can form part of a referral spell where they are open concurrently or where the next referral opens within 5 days of the previous referral ending. A spell is closed when the last referral is recorded as closed (using the service discharge date).

Only referrals which fall within the inclusion and exclusion criteria will make up the referral spell. Where a person has other open referrals within the same provider that are not in scope for either of the two pathways, these referrals will not be included in the spell.

Data which utilises the referral spell methodology will be published in a separate CSV file in the Mental Health Monthly Statistics publication and also will be denoted by Metric IDs that begin with “MRS”.

Full details can be found in the methodological change notice here

From November 2024 data onwards, some small changes to the methodology have been introduced. These include: 

1. Incorporating REFERRAL REJECTION DATE and REFERRAL CLOSURE DATE in the referral-spell derivation section where SERVICE DISCHARGE DATE is not available.
2. Depreciated crisis team type codes have been excluded from the CYP metric, excluding historic referrals with a team type of either A03 (Crisis Resolution Team) or A04 (Home Treatment Service).

NHS England is aware that some digital providers are flowing Outcomes data. Currently, the methodology used makes no attempt to account for the differences in these provider’s data.

In these cases, data supplied is often insufficient to derive a consistent Person ID. As a result, the data is not truly reflective of this activity as only instances where paired scores occurred in the same month would be captured. Consequently, the paired score was not necessarily the first and last scores in the persons referral. If a provider captures sufficient data to derive a consistent Person ID, there should be no issue in the data.

Work is ongoing to determine the best way to handle digital provider data within the CYP Outcomes metrics. Any future changes will be pre-announced through a methodological change notice.

The July 2025 publication (released 11 September 2025) represents the next steps in improving reporting on health inequalities. These next metrics include improved reporting of rates (both crude and standardised) which facilitate the comparison of standardised rates at National, Region and ICB level. These comparison metrics provide the absolute difference between the White British group with the other ethnic groups and the absolute difference between the Most deprived 20% (sometimes referred to as CORE20)  and the other 80%. These two metrics are intended to provide a method for which comparisons of inequality can be made between both between geographical areas and over time.

Initially, just the CYP Access metric (MHS95) will include these additional metrics, however, overtime it is expected that this approach will be expanded to include other metrics. Other methodologies are also being explored to improve the ways in which health inequalities are reported and monitored. 

The July 2025 publication (released on 11 September 2025) included data broken down using the Learning Disability Status and Autism Status fields. These data items were added to MHSDS as part of the version 6 changes which were implemented in April 2024. Given these fields have only recently been introduced and no significant usage of the field has been made, there are significant data quality issues associated with this data currently. However, by releasing this data at this point, it is hoped that it will help to both improve the visibility of the fields and subsequently begin to improve data quality.