189 The national echocardiography database of the United Kingdom (NED-UK) pilot study

Background & Aim Large curated datasets of clinical echocardiographic measurements with associated interpretation are scarce. Due to standardised echocardiographic protocols and diverse patient populations, the UK offers a unique opportunity to link echocardiographic data to study patient outcomes through nationwide records of primary care visits, hospitalisations, and mortality. Our aims were to 1) Develop a toolkit for the extraction and anonymisation of clinical echocardiographic data and 2) Implement the toolkit in the first NHS Trust to demonstrate feasibility. Methods The data were collected under the 'EchoVision' project – an expanding resource of echocardiographic data (UK HRA approval 251473). Customisation of database queries were undertaken by Philips (Koninklijke Philips N.V, Netherlands) staff with input from the first author. Data (measurements, standardised interpretive phrases and free-text comments) from consecutive clinical echocardiograms, undertaken at an NHS Trust and reported upon within Philips Intellispace Cardiovascular system, were extracted using the built-in 'Advanced Analytics' functionality in date-ranged portions using the customised queries. Individual extract files (.csv) were formatted in Microsoft Excel using a 'Visual Basic for Applications' script to get a rectangular dataset, before being merged and processed in RStudio (v2023.09.1) using R (v4.2.1). Scans from patients with an age <18 years or missing age were excluded. After creating the unique study identifier for each scan via a one-way hash function (https://www.quickhash-gui.org/), protected health information was removed. Subsequently, the data were transferred to the research team, and stored on secure severs with password protected access by the study research staff. Results Standard operating procedures and training materials were written for the use of Philips software and for data anonymisation (the 'toolkit'). Data from n=19100 echocardiograms performed in 2022 were extracted, with n=291 exclusions leaving n=18809 scans. The median age was 68 [interquartile range 26] years – the age distribution is shown in figure 1. Female patients comprised n=8324 scans (44.3%), males n=10108 (53.7%) and n=377 (2.0%) had their sex unrecorded. 'Transthoracic' scan type represented n=18475 (98.2%), with the remainder being transoesophageal (n=190), stress (n=82), bubble (n=58), and intracardiac (n=4). The number of echocardiograms performed per month varied from n=1314 in April to n=1841 in November (figure 2). Conclusions The NED-UK pilot study demonstrates the ability to develop a functional toolkit for the extraction and anonymisation of clinical echocardiographic report data and the feasibility of large-scale extraction of echocardiographic data. Future aims should be focused on expanding NED-UK to become a multi-site multi-vendor resource for clinical research with the potential for predicting cardiovascular risk across diverse patient populations. Conflict of Interest None