Skip to main content
The Website Services team at NHS England are testing early designs to understand how you find information. Your views will help us check our ideas work. Sign up to take part in our user research.

Prepare and upload a schema file for a Data Access Request Service (DARS) application

Page contents

This guidance explains how to prepare and upload a schema file as part of a Data Access Request Service (DARS) application.

You must provide a schema file as part of your DARS application if you intend to bring research data into the NHS England Secure Data Environment (SDE). 

The schema file describes the structure of your data set and helps us understand how it will be linked to NHS data sets and used for analysis within the SDE.

If you are bringing in multiple research data sets under the same Data Sharing Agreement (DSA), you must provide a separate schema file for each research data set.

Prepare your schema file

When preparing your file, it is important that it is formatted correctly so it can be uploaded without delay.

Name your schema file

There is no required naming convention, but each schema file within the same application must have a unique filename. If you use the same filename more than once, the file will be rejected.

Format your schema file

Your schema file must be:

  • a CSV (.csv) flat file 
  • no larger than 3MB

Use the following format when preparing the file. The schema headers (for example Field Name, Description, Data Type) are fixed and must be used exactly as shown. These headers must not include underscores.

You can download the schema template as an example of how the file should be formatted.

 

Schema headers Description Content and format guidance  Max field length
Field name Include all the research data field names that appear as column headers in your research data set file. You must include a field named UNIQUE_REFERENCE (often referred to as ‘Study ID’).    
 
  • Use only uppercase or lowercase English letters (A–Z), numbers (0–9) 
  • Use underscores (_) instead of spaces
  • Match the column headers used in your research data set file
 100 characters
Description A short explanation of what each field is and how it will be used. This is required to support review and assurance of the data set.   Provide a clear description and purpose for each field  1000 characters
Data Type The type of data contained in the field.  Only Databricks SQL data types (used in the SDE) are accepted. Choose the relevant data type from the list of data types you can use N/A
Data Format Expected format or pattern of the data, for example, a date pattern or a code structure.  

Formats apply only to DateType and TimestampType values 

Use N/A to show that no format applies 
 
Accepted date formats: 

  • YYYYMMDD 
  • YYYY/MM/DD 
  • YYYY-MM-DD 
  • YYYY.MM.DD 

Accepted timestamp format: 

  • DD-MM-YYYY HH:MM:SS

Accepted Decimaltype format:

  • (precision,scale)  that is, where precision = total number of digits allowed and scale = number of digits after decimal point). For example (5,2)
  
Field length The maximum number of letters or numbers the field can contain.   The value entered must be a whole number. Decimal values will be rejected N/A
Nullable Describes if the field can be left blank. Use TRUE if the field can be left blank or FALSE if the field must always contain a value N/A

Data types you can use

The NHS England Secure Data Environment (SDE) uses Databricks to process and analyse data. For this reason, you must use only the data types below.

Accepted data types Definition When to use this data type Example values
BooleanType A logical value representing true or false When this field can only ever have 2 values and represents a yes/no or true/false condition true, false
DateType A calendar date (year, month, day) When you need to store a date only and time of day is not relevant 2026-12-01
DecimalType   A numeric data type which must have fixed precision (the maximum total number of digits) and scale (the number of digits after the decimal place) For numbers where exact precision is critical, that is, not close enough but correct down to the digit.
When you use DecimalType, you’ll need to define the precision (maximum number of digits) and scale (number of digits on the right of the decimal point). For example (5, 2) can support -999.99 to 999.99

For (5, 2): 567.34
For (3, 1): 33.2
For (6, 0): 456372

Maximum precision is 38.  Scale must be less than or equal to precision. It can be set to 0
DoubleType  A double-precision (64-bit) floating-point number For decimal numbers where greater accuracy is needed. DoubleType will store up to 15 digits 3.1415926535897
FloatType A single-precision (32-bit) floating-point number For decimal numbers where your total digit count does not exceed 7  1.532
IntegerType A 32-bit signed whole number For whole numbers when values are expected to stay within standard integer limits approximately ±2 billion 42
LongType A 64-bit signed whole number For whole numbers that may exceed IntegerType limits  92233720368547758
StringType A sequence of characters (text) For text or identifiers such as IDs or codes, even if they contain only numbers Hello world, ID_329423
TimestampType A date and time value (timestamp), typically down to seconds or microseconds For a specific date and time 16-04-2026 14:30:00

Example schema file and matching research data setExample schema showing the required column headers and structure for a schema CSV file, including field name, description, data type, data format, field length and nullable status.

What the image shows

How a schema file defines the structure of a research data set file.

At the top is a table labelled Schema file. The table lists the required fields for a data set and includes the following column headers: Field name, Description, Data type, Data format, Field length, and Nullable.

Example fields shown in the schema include:

  • UNIQUE_REFERENCE, described as a unique identifier linking each record to a participant. Its data type is a StringType, the field length is 20 characters, and it cannot be empty
  • DATE_READING_1, described as the recorded date of the first reading for the study. Its data type is a date and the format is YYYY/MM/DD
  • READING_1, described as the first reading recorded for the study. Its data type is an IntegerType

A note beside the schema states that all column headers in the research data set file must be defined in the schema file.

Below this is a table labelled Research data set file. This shows example data using the same column headers defined in the schema: UNIQUE_REFERENCE, DATE_READING_1, and READING_1. Each row contains a study reference, a date value, and a numeric reading.

For example, one row contains:

UNIQUE_REFERENCE: Study_012321

DATE_READING_1: 2016/01/01

READING_1: 14

Upload your schema file in DARS online

Follow the steps below to upload your schema file:

  1. Log in to DARS online.
  2. Go to list of uploaded documents.
  3. On the uploading documents page, select Yes, then select Continue.

  4. From the document type dropdown list, select Schema, then select Continue.

  5. Enter a short, meaningful reference for your research data set (maximum 20 characters). The reference will be used to name the research data set in the SDE.

  6. Choose the correct CSV file.

  7. Select Upload and Continue.

If you are providing multiple schema files, upload each file separately before submitting your application.

Files should appear in the List of uploaded documents if they have been uploaded correctly.

What happens after submission

Once you submit your schema file, we will review it alongside your DARS application and contact you if anything needs to be amended. 

During the data application process, you will need to tell us who will submit the participant data (cohort file) and the research data set file. Once your application is approved, the person submitting the data will receive separate guidance on how to access the Secure Electronic File Transfer (SEFT) to upload the files.

You can check the status of your application in DARS online.

Details of the full process for preparing and uploading the files are available in the guidance on bringing research data into the NHS England Secure Data Environment (SDE).

Summary

Here are some quick reminders when preparing the schema file.

Do
  • Ensure the file is a CSV and is 3MB or less
  • Ensure that the schema does not contain any record level or participant data
  • Include a field named UNIQUE_REFERENCE that matches the field used in the participant list (cohort file) and the research data set
  • Check that every field in your research data set is listed in the schema
  • Ensure that field names and schema headers are formatted correctly
  • Ensure that only accepted data types are used
  • Ensure only fields marked Nullable = TRUE are left blank
Don't
  • Include any personal and/or confidential information in your data

Contact us

If you need any support with your request for data, contact us on 0300 303 5678, or email [email protected].

Last edited: 29 October 2026 4:18 pm