Prinicpal Investigator: Christina Saville

Team:

Paul Meredith, University of Southampton

Chiara Dall'Ora, University of Southampton

Tom Weeks, Portsmouth Hospitals University NHS Trust

Sue Wierzbicki, Portsmouth Hospitals University NHS Trust

Peter Griffiths, University of Southampton

Ian Dickerson – Patient and Public Involvement Representative

Start Date: 1 September 2023

End Date: 30 September 2024

Plain English Summary of Findings

Using information about patients already held by hospitals (such as patient demographics, diagnostic information and movements between wards) we estimated the number of nurses needed on the ward each shift. We found that our estimates matched closely with the currently widely-used approach.

For that approach, the nurse in charge records the severity of each patient's illness, and how dependent they are on nursing care, every day or shift.

In contrast our approach uses a type of regression (a tool for finding patterns in data) to automatically calculate the number of nurses needed. This would potentially save nurses time in assessing patients by using information that is already recorded.

What's next?

We used data from one hospital so need to find out if results are similar for other hospitals. We also need to find out whether our estimates relate to patient outcomes.

We have funding for another 1-year study (PREDICT-NURSE validation and extension) to explore this using existing data from another hospital. We will  also investigate whether we can use similar methods in other settings outside acute care, e.g. mental health and community settings.

We have also received funding for a 2.5-year study (PREDICT-NURSE) with at least 5 hospital Trusts collecting new data to develop and test algorithms using a wider range of data and outcome measures. This study will be based on user-centred design, with a national survey and workshops to gather nurses’ and other stakeholders’ views.

Background

Having enough nurses to care for patients on hospital wards is critical for patient safety, but it is difficult to plan for varying numbers of patients and unknown trajectories of deterioration and recovery.  Tools for assessing patients’ needs to help with staff planning are an extra nursing task, thus adding further to workload. We do not know whether ward-level demand could be accurately predicted using existing assessments and data that is already recorded electronically.

The overall aim of the project was to explore the feasibility of predicting acuity/dependency-based workload measures, as assessed by nurses, from routinely collected information in patients’ electronic health records.