A comprehensive high cost drugs dataset from the NHS in England - An OpenSAFELY-TPP Short Data Report

<ns4:p><ns4:bold>Background:</ns4:bold> At the outset of the COVID-19 pandemic, there was no routine comprehensive hospital medicines data from the UK available to researchers. These records can be important for many analyses including the effect of certain medicines on the risk of severe COVID-19 outcomes. With the approval of NHS England, we set out to obtain data on one specific group of medicines, “high-cost drugs” (HCD) which are typically specialist medicines for the management of long-term conditions, prescribed by hospitals to patients. Additionally, we aimed to make these data available to all approved researchers in OpenSAFELY-TPP. This report is intended to support all studies carried out in OpenSAFELY-TPP, and those elsewhere, working with this dataset or similar data.</ns4:p><ns4:p> <ns4:bold>Methods:</ns4:bold> Working with the North East Commissioning Support Unit and NHS Digital, we arranged for collation of a single national HCD dataset to help inform responses to the COVID-19 pandemic. The dataset was developed from payment submissions from hospitals to commissioners.</ns4:p><ns4:p> <ns4:bold>Results: </ns4:bold>In the financial year (FY) 2018/19 there were 2.8 million submissions for 1.1 million unique patient IDs recorded in the HCD. The average number of submissions per patient over the year was 2.6. In FY 2019/20 there were 4.0 million submissions for 1.3 million unique patient IDs. The average number of submissions per patient over the year was 3.1. Of the 21 variables in the dataset, three are now available for analysis in OpenSafely-TPP: Financial year and month of drug being dispensed; drug name; and a description of the drug dispensed.</ns4:p><ns4:p> <ns4:bold>Conclusions: </ns4:bold>We have described the process for sourcing a national HCD dataset, making these data available for COVID-19-related analysis through OpenSAFELY-TPP and provided information on the variables included in the dataset, data coverage and an initial descriptive analysis.</ns4:p>