Validation of Risk Calculator to Predict 30-day Heart Failure Outcomes

  • Research type

    Research Study

  • Full title

    Validation of a Simple Bedside Risk Calculator to Predict 30-day Outcomes Among Patients Hospitalized for Heart Failure

  • IRAS ID

    280209

  • Contact name

    Rajiv Sankaranarayanan

  • Contact email

    Rajiv.Sankaranarayanan@liverpoolft.nhs.uk

  • Sponsor organisation

    Liverpool University Hospitals NHS Foundation Trust

  • Duration of Study in the UK

    0 years, 3 months, 0 days

  • Research summary

    Patients hospitalised for heart failure (HF) are commonly re-hospitalised or die within weeks of discharge, with 30-day readmission rates upto 20-30%. Healthcare utilisation due to hospitalisations and readmissions accounts for more than 70% of the cost of HF care. Risk prediction tools can identify hospitalised patients at high risk of death or readmission so that services can be titrated to risk. Current HF risk-prediction models can reliably predict death but are limited in their ability to predict hospitalisation. Furthermore, risk-prediction tools have limited uptake because they are complex, require the input of multiple variables into regression models and are not designed for use at the point of care. Most HF risk prediction models have been derived and validated using retrospective administrative data.

    Using data from the Patient-Centered Care Transitions in Heart Failure clinical trial, PHRI derived and internally validated a simple point-of-care risk calculator for 30-day risk prediction among 772 patients hospitalised for HF. This model included point-of-care admission N-terminal prohormone Brain Natriuretic Peptide (NT-proBNP) or discharge NT-proBNP, Length of hospital stay, and number of Emergency Department (ED) visits (LENT) in the preceding 6 months. In internal validation studies, the LENT risk score predicts risk of 30-day readmission with equivalent discrimination but greater ease of use than complex risk scores that have been derived using administrative data.
    In this retrospective cohort study, we aim to externally validate the LENT risk prediction model among patients hospitalized for HF at Aintree University Hospital(01-01-2017 to 21-12-2019). We will compare our model against other well-established models: LACE model(Length of hospitalization stay, Acuity of admission, Comorbidities and Emergency department use of patients); the Medicare Readmission Risk Score (RRS model) and the EFFECT score. We hypothesize that the LENT score will demonstrate improved risk discrimination for 30-day readmission compared to existing models.

  • REC name

    West Midlands - Black Country Research Ethics Committee

  • REC reference

    20/WM/0236

  • Date of REC Opinion

    19 Aug 2020

  • REC opinion

    Favourable Opinion