Data quality issues

The CDS UNIQUE IDENTIFIER data item must not include any patient identifiable data, even where included as part of a concatenated set of codes or ‘compound’ code. This represents a potential information governance risk, as this is not anonymised for sensitive records.

Recommended practice

The standard for construction of this data item should be followed. In particular it should be noted that the purpose of this data item is to identify a RECORD for use by the Net Update protocol when updating SUS with a new version of that record.  Please refer to the NHS Data Dictionary for more information.

NHS NUMBER is essential in making it possible to share patient information safely, effectively and accurately across NHS organisations. Where NHS Number is not populated correctly there is a major impact on tracing patients, linking data and deriving additional data items from PDS

Recommended practice

NHS Number must be submitted except for sensitive records, patient objections or those covered under Section 10 of the Data Protection Act.  Please refer to the Data Quality Dashboards for information on counts of errors.

If data is withheld, data senders must populate data item WITHHELD IDENTITY REASON.

The Provider Code (ORGANISATION CODE (CODE OF PROVIDER)) is a mandatory data item. Interchanges are rejected where it contains an invalid code.

Retired or unregistered codes, or codes submitted using non-UPPERCASE (e.g. AbC), or non-alphabetic characters (e.g. A/C) are identified as invalid.  Failure to correctly populate Provider Code will mean that the activity is not accepted for PbR processing, which may lead to payment not being received. Such records will not appear in the Provider’s extracts from SUS.

Recommended practice

Provider code should always be populated with a valid, current organisation code, in the correct format.

Commissioner Code (ORGANISATION CODE (CODE OF COMMISSIONER) is a mandatory data item and an interchange will be rejected if it is omitted.

The majority of invalid Commissioner Codes are out of date or invalid organisation codes.  An unpopulated Commissioner Code could result in the Provider receiving no payment. Such records will not appear in the Commissioner’s extracts from SUS.

Recommended practice

Commissioner Code should always be populated with a valid, current organisation code.  Please refer to the SUS Data Quality Dashboards for information on counts of errors.

CDS ACTIVITY DATE should be populated with the 'originating date' for the activity within the CDS, as described in the NHS Data Dictionary (for example Episode End Date in APC, Appointment Date in OP etc.). Where this is not the case, it is possible that an incorrect Age at Activity Date may have been assigned.  This means that derivations from PDS may not be using the correct time period.

Recommended practice

CDS ACTIVITY DATE should always be populated with the originating date as described in the NHS Data Dictionary.

Where time data items in the A&E CDS are not submitted or the submitted data is not correct to the minute, the usability of the data becomes limited for measuring waiting times in A&E and also potentially compromises use when monitoring breaches of the four-hour wait target.

Recommended practice

Times should always be submitted, as accurately as possible. Please refer to the SUS Data Quality Dashboards for information on counts of errors.

Where SITE CODE (OF TREATMENT) codes are unpopulated, invalid or inaccurate it is not possible to determine the hospital site where treatment took place. This could potentially impact on monitoring of patient safety and limits the use of the information for patient choice.

Recommended practice

SITE CODE (OF TREATMENT) should always be populated with a valid organisation site code. The code should not end in '00' as this gives no information on site, and all records for a Provider should not be submitted under the same site code.

Please refer to the SUS Data Quality Dashboards for information on counts of errors.

Should additional site codes be required, these should first be registered with the Organisation Data Service (ODS).

Morphology codes are 8 characters long and exceed the 3-6-character length validation for Diagnosis data items in the CDS schema. Therefore, Morphology codes fail validation.

In some cases, users have tried to shorten (truncate) the code so that it will be accepted. However, doing so can invalidate the code and result in an undefined HRG (UZ01Z) being derived.

Recommended practice

Morphology codes are for local use only and should not be submitted to SUS.

A&E clinical codes (Diagnosis, Investigation and Treatment) are six-character codes concatenating a number of codes, constructed in specified positions of the code.  Inconsistency in the submission of each part of the codes can lead to them being misinterpreted.

Recommended practice

Standards should be followed for the submission of these codes. See NHS Data Dictionary.

Unpopulated, invalid Diagnosis Codes, use of the 'R69X' (not specified) or retired codes, will mean that records are not able to be ‘grouped’ in SUS PbR. Secondary use of the data is also compromised.

Recommended practice

Valid diagnosis codes should be submitted for all records

In cases where OPERATION STATUS CODE indicates that a procedure took place, but no Procedure Code has been submitted, it is impossible to determine what happened to the patient.  This limits the use of the data and could have financial implications in PbR.

Recommended practice

Both operation status and procedure code data items should be populated with a valid, consistent code.  If no operation took place, operation status should be populated with code ‘8’ – No operation took place.  Please refer to the SUS Data Quality Dashboards for information on counts of errors.

In some cases ICD-10 codes are incorrectly sent in place of A&E diagnosis codes in the A&E CDS with DIAGNOSIS SCHEME IN
USE populated with ‘01’ (Accident & Emergency Diagnosis).

This is not the correct way to flow the data and can impact on A&E diagnosis analysis across areas that include organisations submitting the data incorrectly.

Recommended practice

Organisations should only submit diagnosis scheme in Use ‘01’ (Accident & Emergency Diagnosis) with A&E diagnosis codes. A&E diagnosis codes are required for grouping purposes. ICD-10 and READ codes are optional for additional use in A&E CDS.

Resubmitted records containing a new CDS INTERCHANGE SENDER IDENTITY code do not overwrite the corresponding, previously submitted records because the key data items do not match.

Recommended practice

Users are encouraged to use the ‘Data Quality Dashboard’ and ‘Monthly Database Counts’ to monitor duplicates.
 

It is possible for data submitted in a previous interchange to be lost. This can be due to the following possible reasons:

• Data senders using BULK update submitting General CDS types (120,130,140) separately.  These should all be sent under a single Bulk Replacement Group. Failing to do so will result in records of the latest CDS type overwriting all records in the previous CDS types for the same reporting period.
• Data senders mixing BULK and NET interchanges. Without an advanced understanding of the two update protocols, mixing BULK and NET can lead to these submissions overwriting each other.  It is therefore not recommended to mix BULK and NET.
• Data senders using incorrect BULK Start and End dates in submissions.

More information about Net and Bulk submission methods can be found in the Submitting CDS Data to SUS guidance document. Data senders are also advised to check the Monthly Reconciliation report to monitor counts.

Data senders are also advised to check the Monthly Reconciliation report to monitor counts.

Maternity Data (CDS types 120 or 140) is often submitted as General Episodes (CDS type 130). Where this occurs, these records will be 'cleaned' in HES to become Maternity records, by looking at the Procedure code or DOB/Episode Start Date/ Admission Method/ Episode Order, which will relate to maternity.  However, this will leave blank maternity data items that are mandatory in SUS because they are not available in General Episodes (CDS type 130).

Recommended practice  

Delivery and birth episodes should be submitted as the correct CDS type. Please refer to the Maternity Data Quality Dashboard for numbers of general records that HES will ‘clean’ to Maternity. The Commissioning Data Set Mandated Data Flows of the NHS Data Dictionary lists the mandatory CDS flows.
 

Maternity ‘tails’ are blocks of repeating delivery (birth) information caused by flat- files output by local systems generating extra local groups.

Up to 9 delivery events can be generated this way; forming a Maternity ‘tail’ that contains extraneous activity information. When these records are created, values for the first and only baby can be populated through the extraneous ‘tails’ in the CDS submission. This can create data that appears to represent multiple identical babies for a single delivery.

Recommended practice

Data senders are advised to put the necessary processes in place to ensure that these extra XML data items are not being populated. This would include ensuring that the number of submitted XML groups is equal to the number of babies in the delivery episode.
 

The Outpatient CDS (020) is a retrospective data set and contains information on activity that has happened. The Future Outpatient CDS (021) is a prospective data set containing planned activity.  Where Future Outpatient appointments are submitted as part of the Outpatient (where ATTENDED OR DID NOT ATTEND = 0 - Not applicable - APPOINTMENT occurs in the future) it results in poor data quality in a number of data items relevant to future appointments.

Recommended practice

Future appointments should always be submitted as CDS 021 Future Outpatient and not CDS 020.

When a valid HRG has not been successfully derived in SUS, no tariff can be applied.  This failure may be due to key data items, such as Diagnosis or Procedure, either not being populated where required or being populated with an invalid code.

Recommended practice

Refer to the SUS Data Quality Dashboard to determine if the number of ‘UZ01Z’ (Undefined HRG) codes is higher than expected and the Error Extracts, available via the SUS Portal for detailed reasons for these errors.