Thank you for your request for information about the following:

Request

You asked us:

Since contacting you about this, we have seen this incredibly interesting publication Estimated prescribing patterns for care home patients aged 65 years and over (shinyapps.io) which we found here Data Science projects | NHSBSA

We would like to request access to the underpinning data for this work:

• We would like to request number of prescriptions and cost of prescriptions

• Split by a combination of: age group, sex, care, deprivation, home residence flag and BNF chapter.

A long form extract would be ideal.

(The application won’t currently open for me, but from memory these were all included in the analysis, however, happy to be corrected, particularly wrt deprivation).

Can you let me know if this is possible and what is the most efficient route to obtaining this?

The NHS Business Services Authority (NHSBSA) received your request on 17 October 2024.

We requested clarification on 8 November 2024 with the below:

Question 1

Please can you define what you mean by Home Residence Flag?

Question 2

You have asked for total prescriptions. However, our data is at item level. Additionally, grouping by BNF chapter is requested. Since a prescription may contain items from multiple BNF chapters, this would effectively end up counting some prescriptions more than once (the grouping would be "count of prescriptions with at least one item from a specific BNF chapter"). Therefore, total item count is the advised aggregation. Would this meet your needs?

Question 3

Please can you confirm the Time period, with a start and end date, you would like this information for?

Question 4

Would you like the data aggregated monthly or annually? Please be aware that monthly aggregation is more likely to result in item counts less than 5, which would lead to value suppression for Statistical Disclosure Control (SDC) reasons.

Question 5

Please could you also clarify what you mean by “long form extract”?

You responded with the below on 11 November 2024:

We are seeking the data underpinning this product:   https://nhsbsa-data-analytics.shinyapps.io/estimated-prescribing-patterns-for-care-home-patients/

Question 1

A: some of the charts (eg the one titled “Estimated prescribing metrics by prescribing setting for patients aged 65 years and over in England” show data for care home and non-care home patients. From this we have inferred that you have the data for both care home residents and non-residents, and could provide separate data. Eg values could be: Not in a care home, residential home, nursing home

Question 2

Yes please

Question 3

For the period covered by the report (2020/2021, 2021/22, 2022/23)

Question 4

Please can we have annual and monthly

Question 5

A data table which includes all combinations of the variables and the number of the items and cost vales of each combination. A patient count would also be helpful.

We have handled your request under the Freedom of Information Act (FOIA) 2000.

Our response

I can confirm that the NHSBSA holds the information you have requested and a copy of the information and notes explaining it is attached.

Please read the below notes to ensure correct understanding of the data.

Prescribing Data

Analysis is based on primary care prescription data collected by the NHS Business Services Authority. The data are collected for the operational purpose of reimbursing and remunerating dispensing contractors for the costs of supplying drugs and devices, along with essential and advanced services, to NHS patients.

Prescribing Exclusions

This data excludes:

• Items not dispensed, disallowed and those returned to the contractor for further clarification.

• Prescriptions prescribed and dispensed in Prisons, Hospitals and Private prescriptions

• Items prescribed but not presented for dispensing or not submitted to NHS Prescription Services by the dispense

Unnaccounted Prescribing

Patients may receive prescription items that have not been prescribed to them personally and will not be accounted for. This may occur in the case of high-volume vaccines such as flu vaccines, and in the case of bulk prescribing of products.

Prescribing Reason

The NHSBSA do not capture the clinical indication of a prescription and therefore do not know the reason why a prescription was issued, or the condition it is intended to treat. Many drugs have multiple uses, and although classified in the BNF by their primary therapeutic use may be issued to treat a condition outside of this.

Prescribing Innacuracies

Due to manual processes involved in the processing of prescriptions there may be inaccuracies in capturing prescription information which are then reflected in the data. NHS Prescription Services have a variety of validation streams throughout prescription processing to support accurate capture of the data. In addition, a retrospective sample is completed in the month following reimbursement to identify the accuracy of prescription processing information. The check includes the accuracy of prescriber, practice, and drug information, but does not include the personal details of the patient. The reported Prescription Processing Information Accuracy for the 12-month rolling period ending July 2024 was 99.9%. The sample may not be representative at a more granular level; as such the level of accuracy is undetermined for specific groups such as drugs, geographies, and time periods. It should also be noted that the identification of errors in the accuracy checking sample does not result in amendments to data held in NHSBSA systems.

Further details of Prescription Processing Information Accuracy can be found on The NHSBSA website: https://www.nhsbsa.nhs.uk/pharmacies-gp-practices-and-appliance-contractors/payments-and-pricing/how-we-process-prescriptions Patient Address

The analysis required that every prescription form had a patient address recorded. Addresses were available for all electronic prescriptions. Address information was not captured directly from paper prescriptions and therefore a process was derived to generate these addresses using a mix of information from the Personal Demographic Service (PDS) and electronic prescriptions across a range of months. Although accurate, this is not as robust as directly sourced patient address information from electronic prescriptions.

For the 2022/23 financial year, patient addresses could be allocated for 99.9% of paper prescription forms where the patient’s NHS number could be identified, and the patient was aged 65 years and over. Prescription forms with known non-English patient address information were removed from the analysis. Records with an unknown or missing postcode were included.

Patient NHS Number

The analysis only includes patients with an NHS number and date of birth verified by PDS. NHS numbers are captured for 100% of electronic prescription messages. We estimate that NHS numbers are captured for 94.7% of paper prescriptions. Overall, the capture rate is over 99% of all prescriptions.

The NHSBSA periodically investigate the accuracy of NHS numbers captured from paper forms. The personal details captured (NHS number, date of birth and age) are compared against those on the prescription form for a random sample of 50,000 prescription forms. The NHS number captured typically matches that on the prescription form for over 99.9% of forms. The results represent the accuracy for all prescription items processed; as such the level of accuracy is undetermined for specific medicines, geographies, time periods and other factors. By contrast, the accuracy of captured NHS numbers in electronic prescribing is estimated to be 100%.

Classifying Prescription Forms by Care Home Status

Prescription forms do not indicate whether or not a patient resides in a care home. A methodlogy was developed to categorise prescription forms by their care home status. This resulted in prescription forms being categorised as either a care home or non-care home. The metholodgy can be found here: https://rpubs.com/nhsbsa-data-analytics/methodology

Central to the methodology was the use of an internally developed address matching algorithm. More information on this algorithm can be found here: https://github.com/nhsbsa-data-analytics/addressMatchR

The methodology and algorithm was used to match patient addresses against address records from Ordnance Survey (OS) AddressBase. OS AddressBase is a comprehensive list of UK addresses, which also contains a building classiication. This building classification was used to identify which properties were care homes. More information on OS AddressBase can be found here: https://www.ordnancesurvey.co.uk/products/addressbase

As part of the methodology is dependent upon fuzzy-matching using the above address matching algorithm, and the output is grouped by care home status, these values are to be taken as estimates.

IMD Decile Attribution and Exclusions

An Index of Multiple Deprivation (IMD) decile can be attributed to each Lower Super Output Area (LSOA). This information can be found at: https://www.gov.uk/government/statistics/english-indices-of-deprivation-2019

An IMD decile was then able to be attributed to prescription records using the patient postcode. This was through using the National Statistics Postcode Lookup (NSPL) which attributes an LSOA per postcode. This information can be found at: https://geoportal.statistics.gov.uk/

If a patient postcode was missing or invalid, an IMD decile could not be attributed to a record. Alternately, if a patient postcode was recorded incorrectly, an incorrect IMD could be attributed to a record. As the data was requested to be aggregated by IMD decile, along with several other grouping variables, records without an attributed IMD decile have been excluded from the analysis.

An important point to note for care home prescribing is that the attributed IMD decile was based upon the care home postcode, as this was recorded as the patient's residence. It is possible the IMD of their residence before moving into a care home was located within a different IMD decile.

This is in contrast to non-care home patients, whose attributed IMD decile will be in relation to their 'regular' residence. Meaningful IMD-level comparisons between care home and non-care home patients are therefore not possible.

Statistical Disclosure Control (SDC)

Due to the matching methodology used to classify prescription records by their care home status, output values are considered to be estimates. As a result, aggregated output values have been rounded, as they cannot be considered as being exact.

Any item count, cost value or patient count less than five and greater than zero have been rounded to five. All other rounding is to the nearest ten. It is not possible to differentiate between actual value of five and values that have been rounded to five in the output data.

Tallying or Counting Output Values

Tallied item counts and costs cannot be considered as exact values. This is due to values being estimates and these estimates then being rounded. If values within the grouped data are added together, any resulting value from such a calculation would only be indicative.

Patient counts cannot be tallied. This is because a patient can be counted multiple times across groups. For example, a patient could receive both care home and non-care home prescribing within the same month. Or, if a patient received prescribing from multiple BNF chapters, they would again be double counted if the chapter-level patient counts were added together.

An overarching caveat to all patient counts, is that they only account for patients receiving prescribing. If a care home patient did not recieve prescribing in a given month they would not appear in this data. Patient counts are therefore not of the entire care home population, and only relate to those receiving prescribing.

Care Home Setting

A subset of the data was attributed a care home setting. This was through supplementing the OS AddressBase data with care home-specific data from The Care Quality Commission (CQC). More information on CQC and additional care homes data can be found here: https://www.cqc.org.uk/

It is either not possible or inappropriate to aggregate data with multiple groupings by care home setting. This is because:

• Only a subset of care home records could be attributed a residential or nursing home status.

• Some care home records were classified as both a residential and nurisng home.

• A care home may change between these two types of care home settings within a financial year.

Publishing this response

Please note that this information will be published on our Freedom of Information disclosure log at:

https://opendata.nhsbsa.net/dataset/foi-02316

Your personal details will be removed from the published response.

Data Queries

Please contact foirequests@nhsbsa.nhs.uk ensuring you quote the above reference if you have any specific questions regarding this response; or, if you feel you may be misunderstanding or misinterpreting the information; or, if you plan on publishing the data.

Reusing the data and copyright

If you plan on producing a press or broadcast story based upon the data please contact communicationsteam@nhsbsa.nhs.uk. This is important to ensure that the figures are not misunderstood or misrepresented.

The information supplied to you continues to be protected by the Copyright, Designs and Patents Act 1988 and is subject to NHSBSA copyright. This information is licenced under the terms of the Open Government Licence detailed at:

http://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/

Should you wish to re-use the information you must include the following statement: “Data Warehouse, NHSBSA Copyright 2024” Failure to do so is a breach of the terms of the licence.

Information you receive which is not subject to NHSBSA Copyright continues to be protected by the copyright of the person, or organisation, from which the information originated. Please obtain their permission before reproducing any third party (non NHSBSA Copyright) information.