{
  "abstract": "Background The National Heart Failure Audit gathers data on patients coded at discharge (or death) as having heart failure as the primary reason for admission. To allow comparison of outcomes between individual hospitals, we developed a model to adjust for differences in the baseline risk of the patients.Methods The risk model was developed using logistic regression with robust SEs used to account for clustering of patients within hospitals. All first admissions between 1 April 2017 and 31 March 2018 were used to predict 30-day mortality from the day of admission. We used variables widely available in clinical practice that are known to be associated with prognosis and are independent of quality of care (eg, pharmacotherapy). Temporal validation was performed by applying the risk model to data in the audit data from 2018 to 2019 and 2021–2024. Geographical validation was done by dividing the development dataset according to hospital location.Results Data for 54 080 patients were available for model development and 43 942 patients for the 2018–2019 validation cohort. Ten variables contributed to the model, which showed good discriminatory ability with a C-statistic of 0.80 (95% CI 0.79 to 0.81). Calibration slope was 1.00 (95% CI 0.97 to 1.03) and calibration-in-the-large −0.02 (95% CI −0.06 to 0.01). The observed and predicted mortality showed good agreement across all deciles of risk. Model performance for subsequent years was similar. Geographical validation was similarly satisfactory.Conclusion A risk model based on a few widely available variables accurately predicts mortality within 30 days of hospital admission for heart failure, which may enable fair comparisons between hospitals in England and Wales.",
  "authors": [
    {
      "affiliations": [
        "Cardiology, Hull University Teaching Hospitals NHS Trust, Hull, UK"
      ],
      "name": "Andrew L Clark"
    },
    {
      "affiliations": [
        "Biostatistics Group, University College London, London, UK"
      ],
      "name": "Rumana Z Omar"
    },
    {
      "affiliations": [
        "Department of Statistical Science, University College London, London, UK"
      ],
      "name": "Gareth Ambler"
    },
    {
      "affiliations": [
        "MAC Clinical Research, Manchester, UK"
      ],
      "name": "Chen Qu"
    },
    {
      "affiliations": [
        "School of Cardiovascular & Metabolic Health, University of Glasgow, Glasgow, UK"
      ],
      "name": "John G Cleland"
    },
    {
      "affiliations": [
        "NHS Arden & GEM Commissioning Support Unit, Leicester, UK"
      ],
      "name": "Aminat Shote"
    },
    {
      "affiliations": [
        "National Institute for Cardiovascular Outcomes Research, Leicester, UK"
      ],
      "name": "Peter D Jones"
    },
    {
      "affiliations": [
        "National Institute for Cardiovascular Outcomes Research, Leicester, UK"
      ],
      "name": "Jiaqiu Wang"
    },
    {
      "affiliations": [
        "Institute for Cardiovascular Outcomes Research, NHS Arden and Greater East Midlands Commissioning Support Unit, Leicester, UK"
      ],
      "name": "Mark A de Belder"
    },
    {
      "affiliations": [
        "Clinical & Academic Department Cardiovascular medicine (3A), Whittington Hospital, London, UK",
        "University College London (Hon), London, UK"
      ],
      "name": "Suzanna Marie Hardman"
    },
    {
      "affiliations": [
        "Department of Cardiology, King’s College London, London, UK"
      ],
      "name": "Theresa McDonagh"
    }
  ],
  "title": "Predicting 30-day mortality after hospital admission for heart failure: the National Heart Failure Audit for England and Wales",
  "uid": "004e84f4-cd02-5d1e-bda5-0ec6288da963"
}