Learning disabilities health check scheme: supporting information

The learning disabilities health check (LDHC) scheme is a GP enhanced service. The monthly publication includes a count of patients who have a QOF diagnostic learning disability, and indicators concerning the proportions of these patients who have received a learning disability annual health check, and who have completed a health action plan.

GP Contract Services form part of the General Medical Services (GMS) contract, which covers the delivery of primary care services across England, agreed between NHS England and the British Medical Association's (BMA) General Practitioners Committee (GPC). These data are primarily used for payment and management information purposes as well as for wider usage to help support commissioning, planning and policy decisions.

From the 2022-23 service year onwards, the LDHC service has been changed. The format of the payment count (relating to the number of patients who have received an annual learning disabilities health check) is unchanged, but all other counts have been replaced with fractional indicators, which are comprised of numerators, denominators, personalised care adjustments and exclusions. 

For more information on how these indicators are constructed, please see the 'Using the .csv file' section of this page, and for more detail review the business rules documents. 

The components of these indicators are defined in the data dictionary section of this page.

Suppression has been applied to the .csv file where small numbers could be disclosive. Suppressed figures are represented with *, and the logic for the suppression is as follows:

Where the denominator count for an indicator < 2 and the value of one PCA count for the indicator is equal to the sum of all PCA counts for an indicator, all PCA values and the value of the denominator are suppressed.

The impact of this upon the accuracy of the total counts is small; for example in November 2022, 33 practices' (0.5%) LDHCMI034 indicators were subject to suppression.

Note: for the period April 2022 - October 2022, the logic used was slightly different:

Where the denominator count for an indicator = 0 and the value of one PCA count for the indicator is equal to the sum of all PCA counts for an indicator, all PCA values and the value of the associated register are suppressed.

The impact of this upon the accuracy of the total counts was also small; for example in July 2022, 4 (0.03%) registers were subject to suppression.

The .csv file is structured as follows with example data items provided:

The accompanying data dictionary file provides descriptions of the indicators and measures included in the .csv file, and can be joined to the .csv file on the indicator code (IND_CODE) and measure (MEASURE) fields respectively. Examples of the data dictionaries are provided below to show how they are structured:

Indicators dictionary

Measures dictionary

The learning disabilities health checks data are released in monthly .csv files. These files can be loaded into a database for querying, or can be interrogated using the pivot tables feature in Microsoft Excel.

Items included

The .csv file includes the following items:

Counts

These are comprised of a single measure, identified in the data dictionary as items with a MEASURE of 'Count'.

Fractional indicators

These are comprised of a number of measures:

• Numerator - the number of patients who have received the relevant intervention. Identified in the data dictionary as items with a MEASURE of 'Numerator'.
• Denominator - the number of patients who were eligible to receive the relevant intervention. Identified in the data dictionary as items with a MEASURE of 'Denominator'.

• Personalised care adjustment (PCA) - the number of patients excepted from the indicator due to having a personalised care adjustment (PCA) specified.
  
  PCAs can be applied to patients for a number of specified reasons and are usually the result of a patient or a GP decision at a personal level. Examples of PCAs could be patient or carer refusal of treatment, a patient cancels or does not attend a consultation appointment, or a GP’s advice that two types of medication or treatment methodology should not be administered simultaneously.
  
  Patients with PCAs are not included in indicator denominators.  However, where a patient has received the intervention and also has a PCA recorded against an indicator, the patient will be counted in the numerator and the denominator but not against the PCA (i.e. intervention trumps PCA).
  
  PCAs can be identified as measures with a MEASURE_TYPE of 'PCA' in the data dictionary.

• Exclusions - the number of patients who are not eligible for inclusion in the denominator for definitional reasons (e.g. does not fall within the specified age range). Identified as measures with a 'MEASURE_TYPE' of 'Exclusion' in the data dictionary.

Data quality measures

There are three data quality measures, identified in the data dictionary as MEASURE DQ_FLAG_1, DQ_FLAG_2, and DQ_FLAG_3. All show that the identified practice/indicator combination has a data quality issue. Descriptions of the different measures can be found in the Data Quality section below.

Calculations

The following information can be derived from the .csv file:

Payment count

The monthly payment count (LDHC021) can be found against the measure ('Count') associated with the indicator code.

Calculating the learning disability register

The learning disability register (patients with a learning disability as per the QOF definition) can be calculated by summing the measure ('Count') associated with the two age-specific register items LD_REG_1417 and LD_REG_18OV:

Register = LD_REG_1417 + LD_REG_18OV

Calculating indicators

The learning disabilities health check scheme includes fractional indicators. The indicators take no account of “PCAs” (patients to whom the indicator applies, but who are not included in the indicator denominator according to agreed PCA criteria).

Indicators can be calculated according to the formula:

Indicator \(= ({{Indicator Numerator} \over Indicator Denominator})* 100          \)

Calculating the percentage of patients receiving the intervention

The percentage of patients receiving the intervention gives a more accurate indication of the rate of the provision of interventions as the denominator for this measure covers all patients the indicator applies to, regardless of PCA status (i.e. indicator PCAs and indicator denominator). This is calculated for an indicator as follows:

Percentage of patients receiving the intervention \(= ({{Indicator Numerator} \over (Indicator Denominator + Indicator PCAs)}) * 100          \)

Calculating personalised care adjustment (PCA) rates

The PCA rate for an indicator can be calculated according to the formula:

PCA rate \(= ({{Indicator PCAs} \over {Indicator PCAs + Indicator Denominator}})* 100          \)

Coverage

The coverage of the dataset is reflective of the number of practices that have participated in the service. Typically, participation increases throughout the reporting year; therefore the number of practices included in the publication tends to be lower in April and increases month-on-month.

Data quality checks

The following checks have been carried out, and are identified by way of data quality flags in the .csv file:

Flag 1

The data quality flag representing a numerator greater than a denominator within an indicator. If Numerator > Denominator, DQ_FLAG_1 is included as a measure with value 1 to the respective practice and indicator. 

Flag 2 [decommissioned December 2022]

The data quality flag representing that there is missing information. If one or more expected measure is not present, DQ_FLAG_2 is added as a measure with value 1 to the respective practice and indicator. 

Flag 3

The data quality flag representing that the payment count (LDHC021) is greater than the total count of patients on the learning disabilities register. If LDHC021 > LD_REG_1417 + LD_REG18OV, DQ_FLAG_3 is added as a measure with value 1 to the respective practice.

Incomplete extracts

From January 2023, if one or more expected measures is not present in the data extracted for a practice, that practice will not be included in the dataset. This typically affects 0 to 2 practices per month.