1. Introduction#

Data about mental health services was introduced in April 2000 to provide clinical information to align with the requirements of the National Service Framework for Mental Health. Published in 1999 by the Department of Health, this framework set a 10-year agenda for improving mental healthcare for working-age adults in England (for further information see Government policy and the National Service Framework for Mental Health: modelling and costing services in England). Over the years, the mental health dataset has evolved significantly as summarised in Figure 1. The first publication of the data was released in 2008, covering the period from 2003 to 2007, and was titled the Mental Health Minimum Dataset (MHMDS). MHMDS was replaced in 2014 by the Mental Health and Learning Disabilities Data Set (MHLDDS), which in turn was replaced in 2016 by the Mental Health Services Data Set (MHSDS).

Figure 1 Evolution of NHS England datasets from 2003 to the present day

Table 1 Key characteristics of each of the mental health datasets

2. Strengths of MHSDS#

• MHSDS is designed for administrative purposes; however, it provides comprehensive coverage of NHS-funded mental health, learning disability, and autism services, promoting a better understanding of mental health conditions.

• The dataset tracks individuals over time, enabling researchers to study care pathways, continuity of care, and long-term conditions.

• It includes clinical fields such as diagnoses, interventions, risk assessments, and legal status, which support clinical and epidemiological analyses.

• The dataset captures a wide range of community-based contacts and interventions, allowing research that reflects real-world care.

3. Limitations of MHSDS#

• Certain data fields are missing from MHSDS because these details may not be effectively captured, or it may not be mandatory for providers to submit them.

• Gaps can occur in understanding a person’s full care pathway, as not all voluntary-sector, privately funded, or local authority–commissioned mental health services submit data.

• Diagnosis and treatment details might be recorded late or inconsistently, making some fields difficult to use for research. Unlike other datasets, MHSDS captures data over a shorter time period, which limits long-term analyses.

• Very little diagnostic information can be obtained from the MHSDS diagnosis table. To gain a deeper understanding of mental health conditions, this table must be linked to other datasets — such as GP data — because MHSDS contains relatively few conditions. However, access to GP data is currently only permitted for COVID-related research.

• Although the dataset was commissioned in 2016, it includes records that predate this period. This may introduce inconsistencies in data quality and completeness, which researchers should consider when conducting analyses.

4. Scope and coverage#

The MHSDS collects information about patient activity and care within inpatient services and community services. It also records specific service areas, including the Learning Disability Census and Assuring Transformation Collection, Eating disorders in Children and Young People, Problem Gambling Services or gambling disorders and Early intervention in psychosis. While these services are mainly delivered in community settings, they may also involve inpatient care for patients in a crisis state. Inpatient services in scope include acute inpatient care, psychiatric intensive care, learning disability and autism inpatient services care, specialist inpatient care, and forensic and secure inpatient care. Community services in scope include organic and functional community mental health teams, crisis and liaison services, Children and Young People’s Mental Health Services (CYPMHS), specialist community services, and triage and Single Point of Access (SPA) services.

The MHSDS includes information about patients who are attending a mental health service in England. If the person is wholly or partially funded by the NHS, data submission for that person is mandatory. If the person is wholly funded by any means that is not NHS, data submission for that person is optional. For each person attending a service located outside England but commissioned by an English Clinical Commissioning Group (CCG) or more recently an Integrated Care Board (ICB), data submission for that person is optional.

The MHSDS does not include information about patients who receive mental health care but are not diagnosed with a mental illness, learning disability, or autism spectrum disorder. Additionally, the MHSDS does not capture data on patients receiving treatment for conditions such as anxiety and depression from psychological wellbeing practitioners or counsellors, where services are not delivered through secondary mental health care. Although the MHSDS includes some information on anxiety and depression, the level of detail is limited and not adequate for detailed analysis. More comprehensive information about anxiety and depression is available in the NHS Talking Therapies for Anxiety and Depression (IAPT) dataset.

5. Data collection methodology#

NHS England collects MHSDS data through monthly submissions from service providers via the Strategic Data Collection Service (SDCS) Cloud. The patient-level information is extracted by the providers and commissioners from the local systems and uploaded into the cloud system. The process includes validation checks, and under the Multiple Submission Window Model, providers can resubmit corrected data within the reporting cycle. Earlier versions used the Bureau Service Portal, but SDCS Cloud replaced it to simplify submission and remove the need for special NHS networks.

6. Structure of the dataset#

MHSDS is comprised of more than 60 tables or datasets, which are grouped into themes capturing patient demographics, referrals, care episodes, diagnoses (clinically coded classifications and terminology), Mental Health Act (MHA) episodes, hospital provider spells, staff service, group sessions and care contact information. Each category consists of identifiers to enable linkage of tables across the dataset. Below we provide a summary of each category of tables available in the MHSDS.

Table 2 Categories of tables within the MHSDS

7. Coding systems used#

MHSDS uses both ICD-10 codes and SNOMED CT codes, according to the NHS Data Model and Dictionary.

ICD-10 codes are used to represent disease diagnoses and appear in the following tables:

• MHS601 – Previous Diagnosis

• MHS603 – Provisional Diagnosis

• MHS604 – Primary Diagnosis

• MHS605 – Secondary Diagnosis

SNOMED CT codes are used to represent clinical and procedural terms. However, formatting issues such as loss of precision in 16-digit codes and conversion to scientific notation, have impacted their usability. These codes are now defined as numeric values in the NHS Data Model and Dictionary. More information is available via SNOMED International - Service Migration.

Further information about coding systems can be found in the Coded variables guide.

8. Evolution of the dataset#

Over time the dataset has undergone various incremental changes without any major expansion in scope. The current version is V6.0, which was released in 2024.

Table 4 A summary of MHSDS versions between 2015 and the present day

¹ The year of release is the date specified in the user guidance reports, e.g. MHSDS V1.0 was released in 2015, but became mandatory in 2016.

9. Availability in the UK LLC TRE#

The Data Sharing Agreement between University of Bristol (Data Controller of UK LLC) and NHS England stipulates that the following MHSDS packages should flow to UK LLC:

• Package 1d Mental Health Services Community Activity ALL: this package is community activity - admin data, clinical data, demographics.

• Package 2d Mental Health Services Inpatient activity ALL: this package includes everything in 2c plus details of individual incidents of assault, restrictive intervention, self harm and substance misuse.

• Package 3 Mental Health Services All Users: this package provides full information about the people using services, but excludes details of contact with services.

• Package 4 Mental Health Services All Users: this package provides detailed information about mental health currency assessments and assignments.

These packages cover more than 60 MHSDS tables. However, the extract from NHS England that flowed in 2024, only included 33 tables. UK LLC has worked with NHS England to ensure that all tables will flow in the latest extract. This section will be updated with more information once the extract is complete. A detailed overview of the tables available in the UK LLC TRE is provided below, including a pictorial representation of table relationships (Figure 2) followed by a summary table of categories and linkages (Table 3).

Figure 2 Relationships between the MHSDS tables available in the UK LLC TRE

Table 3 MHSDS tables available in the UK LLC TRE, including summary, key variables and linkages

10. UK LLC transformations of the dataset#

All variables which identify organisations (e.g. GP practice, NHS Trust) or geographic areas smaller than a region (e.g. LSOA) are encrypted before being ingested into the UK LLC TRE. The encrypted variables (identifiable by the suffix _e) enable researchers to identify which participants were treated by the same organisation, or live in the same area, but not to identify the organisation or area.

11. Tips for researchers using MHSDS in the UK LLC TRE#

Very few researchers have used the MHSDS in a longitudinal research context. Below are some tips – we will regularly update this page as we gather more information. Please also see the list of mental health FAQs.

1. In MHSDS, the term “referral” does not follow the conventional definition. Instead, it refers to any mental health-related contact with a professional, such as in A&E, a school, a GP practice, with a social worker, within the justice system, or even self-referral.

2. The cohort key and the record number are available within each table (except MHS 301 group session) and so researchers should be able to link between tables.

12. Useful syntax#

Below we will include syntax that may be helpful to other researchers in the UK LLC TRE. For longer scripts, we will include a snippet of the code plus a link to the UK LLC Github repository where you can find the full scripts.

13. Further reading#

Information in this section will be added in due course.