Skip to the content

ADOPTED: Consequences, costs and cost-effectiveness of different workforce configurations in English acute hospitals: a longitudinal retrospective study using routinely collected data

Principal Investigator: Professor Peter Griffiths  

Deputy Principal Investigator: Dr Chiara Dall’Ora 

Professor Jane Ball Co-investigator – nursing workforce, Dr David Culliford Co-investigator - statistics, Dr Jeremy Jones Co-investigator – health economics, Ms Francesca Lambert Co-investigator – patient and public involvement (lay researcher),  Dr Paul Meredith Co-investigator – health informatics, Paul Schmidt Co-investigator – clinical medicine, Talia Emmanuel PhD student, Bruna Rubbo Senior Research Assistant / Research Fellow and Christina Saville Research Fellow 

Partners: Portsmouth Hospitals University NHS Trust, University of Southampton

Started: March 2020

Ends: February 2023

Aim: This study seeks to understand how variation in the size and make-up of care teams on hospital wards in England influences patient outcomes and the costs of care. Background: Research shows that low registered nurse (RN) staffing levels on hospital wards are linked to undesirable outcomes. These include increased poor experiences for patients, an increased risk of dying and, potentially, other outcomes that are bad for patients and increase the cost of care. These include falls, longer stays and unplanned readmissions. For a long time, studies used hospital level averages rather than looking at what happened to individual patients. This uncertainty makes it hard to understand the likely costs and benefits from investing in staff differently. Developments in information technology now make it possible to link nurse staffing levels experienced by individual patients on every day of their stay, to the outcomes experienced by those patients. Our research group was the first to use these new sources of information to explore how the mix of staff in the nursing team affected outcomes and cost of care. We found that each additional hour of RN time per patient reduced the risk of death and shortened their hospital stay. We found that a small reduction in assistant staff, and a small increase in RNs would improve outcomes with no overall increase in costs. Such findings have implications for how hospitals respond to nurse shortages, but the results come from one hospital and use limited costs and outcomes. It is important to see if the conclusions apply more widely. As RNs are in short supply it is also important to better understand how other staff contribute. Design & methods: Our study is in two parts. Study 1 uses information about ward staff and patients' outcomes drawn from hospital electronic systems. We will use anonymous records gathered from all patients and staff in at least 4 NHS Trusts. Using statistical models, we will estimate the impact of RN and assistant staff levels on outcomes. For example, whether the risk of death is lower when more RNs are working on a ward. We will estimate staff costs and also the costs of events such as unplanned readmission or longer hospital stay. We will estimate the cost per 'quality adjusted life year' associated with changes in nurse staffing. Such measures help policy makers to compare the results of investments in health care and put more value on each year where people are expected to be healthy and independent. Study 2 will analyse national data at a whole hospital level to see how the size of other staff groups (e.g. therapy staff and doctors) might influence outcomes. Patient and Public Involvement: Safe staffing in hospitals is an area of public concern. We developed this proposal with this concern in mind, and shaped it through conversations and consultations with patients and members of the public. A member of the public/carer (who is a co-applicant) will facilitate PPI at all stages of the project, with ongoing engagement and sense checking with patients/public to inform analysis, interpretation and presentation of results. Dissemination: Results will be of interest to a diverse audience. We will present findings to national and international conferences and to policy makers, publish in academic journals and present to stakeholders. We will use professional networks and social media to ensure that outputs reach professional, research and public audiences.

Background: The NHS is facing significant challenges in recruiting and retaining staff, particularly registered nurses (RNs). Recruiting unregistered staff is often adopted as a solution to the RN shortage; however, our recent research - the first in England to use longitudinal routinely collected data - found a negative effect of low RN staffing levels on mortality with no evidence that high levels of assistant staff could mitigate the increased risk. Our economic modelling suggested that increases in RN skill mix were potentially cost-effective, but these findings derive from a single NHS hospital Trust with limited cost and outcome data. Aims and objectives: This project aims to estimate the consequences, costs and cost effectiveness of variation in the size and composition of the staff on hospital wards in England. We will build on findings from our previous study, where we looked at staffing on wards in a single hospital. In order to provide estimates that are more likely to apply across the NHS, this study will include at least four hospitals and consider a wider range of outcomes and sources of costs, including death within 30 days of admission, adverse events such as infections, length of hospital stay, readmissions and rates of staff sickness. In order to determine if results are likely to be sensitive to staff groups not on ward rosters we will use national routine data to explore the associations with staffing levels of other groups including medical and therapy staff. Methods: Study 1 will be a retrospective longitudinal observational study with routinely collected data on ward and shift level nurse staffing, and patient outcomes. Data will be derived from the E-Roster systems, used by hospitals to record all planned and worked shifts. We will consider all rostered direct care staff. These data will be linked to patient data derived from the hospital patient administration system (PAS); and other clinical systems and databases of adverse events (e.g. datix). Relationships between RN and assistant staffing levels and outcomes will be explored using survival models incorporating mixed effects. We will use the results of these analyses to model the costs and consequences of different staffing configurations and to estimate the incremental cost-effectiveness associated with change. We will estimate cost per Quality Adjusted Life Year gained or lost (QALY), associated with each staffing configuration using the DANQALE approach. Study 2 will be a panel study using routine national workforce data and outcomes (standardised mortality indicators, patient experience) to consider all staff groups including medical and therapy staff at the hospital level. This study will generate hypotheses about staffing for other groups, confirm the independence (or otherwise) of nurse staffing effects and fill a significant gap in international literature about the association between hospital safety and non-nursing staff levels. Timelines for delivery and impact: our study will be undertaken over 30 months and will provide evidence to inform staffing levels and skill mix planning in the NHS, highlighting potential cost savings, and offering improved patient safety and reduced adverse staff outcomes. To ensure impact, we will work with patients, nurses and key policy makers at all stages; we will publish papers and present to academic and professional conferences, as well as writing lay reports and engaging with traditional and social media