{
  "abstract": "Introduction Echocardiography is widely used to assess left ventricular diastolic function but requires complex guideline-directed assessment. 1–4 We evaluated how diastolic function is reported across NHS district general hospitals and tertiary centres in real-world practice. We then compared findings to guideline-directed assessment of diastolic function generated by a fully automated tool we have developed based on international guidelines, to identify whether there are gaps in current practice.Methods Routinely collected anonymised echocardiography data were extracted from patient records within hospitals forming part of a large academic NHS Trust via the EchoVision observational study (19/HRA/2068), with additional manual extraction from five independent NHS Trusts across England through the ECHO-AID study (20/HRA/1405) to ensure generalisability of results ( figure 1). Data were anonymised locally prior to transfer, and cleaning, merging, and post-processing was performed in R. The dataset includes quantitative echocardiographic measurements and free-text comments, from which operator-reported diastolic function was extracted using bespoke keyword search algorithms. Both American and British Society of Echocardiography diastolic assessment guidelines were coded into an automated pipeline and validated against expert consensus. Operator-reported and automated assessments were then compared. Temporal trends in agreement were analysed using linear regression.Results Data from 150,574 echocardiograms performed between 2017 and 2020 were analysed. Diastolic function was reported in 96,916 (64.4%) studies, of which 84,889 (87.6%) used detailed language to characterise diastolic function (normal, grade I, grade II, grade III, indeterminate, unclassifiable), with 12,027 (12.4%) referencing ‘impaired diastolic function’ or elevated filling pressures without further information. Diastolic function categorisation was achieved in 100% of studies using our automated pipeline; however, 51% of studies were unclassifiable when applying diastolic function guidelines due to confounding factors or missing data. Concordance between operator and automated categorisation increased significantly over time, with the interquartile range of percentage-point differences in categorisation increasing by 1.33 percentage points per year (β = 1.33, p = 0.027), or a total increase of 10.6% since 2015 ( figure 2).Conclusions In a large dataset of routinely collected echocardiographic reports, over a quarter had no diastolic function reported, with similar patterns observed across all hospitals. Our automated diastolic function algorithm was able to categorise all patients and highlighted important differences between human and automated reporting. However, discrepancies have reduced over time. Whether automated tools could improve consistency, speed and accuracy of diastolic function reporting warrants further evaluation.References Nagueh SF, et al. Recommendations for the evaluation of left ventricular diastolic function by echocardiography and for heart failure with preserved ejection fraction diagnosis: an update from the american society of echocardiography. Journal of the American Society of Echocardiography 2020;38:537–569.Robinson S, et al. The assessment of left ventricular diastolic function: guidance and recommendations from the British society of echocardiography. Echo Res Pract 2024;11:16.Mathew T, et al. A Guideline Protocol for the Echocardiographic Assessment of Diastolic Dysfunction. https://www.bsecho.org/common/Uploaded%20files/Education/Protocols%20and%20guidelines/Diastolic%20dysfunction.pdf (2013).Sitges M, et al. EACVI survey on the evaluation of left ventricular diastolic function. Eur Heart J Cardiovasc Imaging 2021;22:1098–1105.Abstract 260 Figure 1EchoVision and ECHO-AID dataset of routinely collected echocardiogram report detailsAbstract 260 Figure 2Temporal trends in percentage point difference between reported and automated diastolic function categorisation",
  "authors": [
    {
      "affiliations": [
        "Cardiovascular Clinical Research Facility, Division of Cardiovascular Medicine, Radcliffe Department of Medicine, University of Oxford, Oxford, United Kingdom"
      ],
      "name": "Casey Johnson"
    },
    {
      "affiliations": [
        "Cardiovascular Clinical Research Facility, Division of Cardiovascular Medicine, Radcliffe Department of Medicine, University of Oxford, Oxford, United Kingdom"
      ],
      "name": "Andrew Fletcher"
    },
    {
      "affiliations": [
        "Cardiovascular Clinical Research Facility, Division of Cardiovascular Medicine, Radcliffe Department of Medicine, University of Oxford, Oxford, United States"
      ],
      "name": "Samuel Krasner"
    },
    {
      "affiliations": [
        "University Hospitals North Midlands NHS Trust, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Sadie Bennett"
    },
    {
      "affiliations": [
        "Oxford University Hospitals NHS Foundation Trust, Oxford, United Kingdom"
      ],
      "name": "Michael Johnston"
    },
    {
      "affiliations": [
        "Worcestershire Acute Hospitals NHS Trust, Worcester, United Kingdom"
      ],
      "name": "Jasper Trevelyan"
    },
    {
      "affiliations": [
        "Worcestershire Acute Hospitals NHS Trust, Worcester, United Kingdom"
      ],
      "name": "Angela Doughty"
    },
    {
      "affiliations": [
        "The Rotherham NHS Foundation Trust, Rotherham, United Kingdom"
      ],
      "name": "Lucy Broadhurst"
    },
    {
      "affiliations": [
        "The Rotherham NHS Foundation Trust, Rotherham, United Kingdom"
      ],
      "name": "Darryl Bradley"
    },
    {
      "affiliations": [
        "Royal United Hospitals Bath NHS Foundation Trust, Bath, United Kingdom"
      ],
      "name": "Daniel X Augustine"
    },
    {
      "affiliations": [
        "Royal United Hospitals Bath NHS Foundation Trust, Bath, United Kingdom"
      ],
      "name": "James Willis"
    },
    {
      "affiliations": [
        "Wrightington, Wigan, and Leigh NHS Foundation Trust, Wigan, United Kingdom"
      ],
      "name": "Haytham Hamdan"
    },
    {
      "affiliations": [
        "St. George’s University Hospitals NHS Foundation Trust, London, United Kingdom"
      ],
      "name": "Jamie O’Driscoll"
    },
    {
      "affiliations": [
        "St. George’s University Hospitals NHS Foundation Trust, London, United Kingdom"
      ],
      "name": "Rajan Sharma"
    },
    {
      "affiliations": [
        "Ultromics Ltd., Oxford, United Kingdom"
      ],
      "name": "Ross Upton"
    },
    {
      "affiliations": [
        "Oxford Centre for Diabetes, Endocrinology, and Metabolism, Radcliffe Department of Medicine, University of Oxford, Oxford, United Kingdom"
      ],
      "name": "Amanda Adler"
    },
    {
      "affiliations": [
        "Cardiovascular Clinical Research Facility, Division of Cardiovascular Medicine, Radcliffe Department of Medicine, University of Oxford, Oxford, United Kingdom"
      ],
      "name": "Paul Leeson"
    }
  ],
  "title": "260 Guideline discordance in real-world human reporting of left ventricular diastolic function",
  "uid": "9be8c73f-f1ae-544d-b42d-0b100ac7e59e"
}