Research Data

The research records contain linked data from a wide variety of health and care settings. For example, the following unit types can contribute to the database:

  • General Practice
  • Child Health
  • Community Care
  • Palliative Hospital
  • Out-of-Hours
  • Urgent Care
  • Accident & Emergency
  • Acute Hospital
  • Social Services

These records are already shared within clinical practice on SystmOne; there is no need for an additional linkage step to produce broad data sets from ResearchOne. Patient flow across primary and secondary care settings can be drawn directly from the research data.

This model also enables contemporary data to be drawn from all sources – the most up-to-date data is available to researchers and there is no need to wait for external data sets to be compiled. For example, projects assessing the impact of primary care initiatives on secondary care do not have to wait several weeks, or even months, for associated community data or Secondary Uses Service (SUS) data to be available to complete the data set.

Data Items

ResearchOne maintains a comprehensive set of well structured data items, drawn from both administrative and clinical data across primary and secondary settings. This includes, for example, diagnostic codes, procedure codes, pathology test data, prescribing data, deprivation indices and care pathways. A full list of stored data items is available in the database summary or as part of a research data access application.

Coded Data

ResearchOne contains coded data from a variety of terminologies, classifications and data dictionaries. This is dependent on the type of data item and the type of organisation who recorded it. For example, there is coded data from:

  • CTV3 (Read codes – version 3)
  • ICD10 (International Classification of Diseases – version 10)
  • OPCS4 (Classification of Interventions and Procedures)
  • A&E Diagnosis, Treatment and Investigation codes
  • DM&D (Dictionary of Medicines and Drugs)
  • British National Formulary (BNF)
  • NHS Data Dictionary

We have expertise in mapping between many different coding schemes; this is often essential if a project involves linking multiple external data sets. For example, we support mappings between version 2 of the Read codes, CTV3 and SNOMED CT, and between DM&D codes and the British National Formulary (BNF).  Please read the codeset information in the Documentation  and Research FAQs sections for further information.

Dynamic Research Data

The ResearchOne database is updated on a weekly basis from new data entered as part of routine practice on TPP SystmOne. This means that researchers have access to up-to-date quality research data at any time. The model also allows for continuous outcome data from any new research-driven interventions to be analysed within an exceptionally short time frame.

Research projects may identify missing data from their research data set which need to be collected in clinical practice. The close links between the clinical system and the research data set are invaluable in these situations. For example, new templates for recording data can be created within the system and distributed automatically to specific organisations. TPP SystmOne also has monthly software maintenance releases if a new data item needs to be collected as part of a larger project.

This means researchers can get access to the new improved data set from ResearchOne in a matter of weeks, without the need for clinicians to enter data in a new, temporary system specifically for the project.

Data Quality Checks

The data held on ResearchOne is centrally checked on a continuous basis to assess data integrity, quality and representation. For example, the data is checked against national prescribing and mortality rates, against aggregate census data and against deprivation and rurality indices.

Details of the relevant reports are available for researchers upon request, with the possibility of bespoke analysis for a new project. Details of any data cleansing or record removal are always supplied to researchers. This may aid with any necessary data imputation, for example, and ensure that any bias is properly accounted for.