- New research funded by NHS England suggests registered nurse numbers linked to improved patient outcomes
- Study questions use of Carter review’s care hours per patient day metric as a safe staffing tool
- Paper concludes staffing link to outcomes is real and can be calculated
An innovative new study into nurse staffing levels at an NHS trust has identified a “calculable” link between nurse numbers and patient outcomes.
A growing body of research around the world has shown an association between nurse numbers and patient care, but this latest study, funded by NHS England, shows the link appears to be real and can be observed and calculated by using high quality patient data.
The research was led by Professor Alison Leary, from London South Bank University, and involved studying 120 million patient data records collected over nine years at University Hospitals Coventry and Warwickshire Trust.
It is the first study of its kind to use “big data” mathematical approaches to staffing. Professor Leary said the results were a first step to showing causation between the number of nurses on a ward and quality of care. Overall the paper identified 40 separate correlations with staffing levels.
She told HSJ: “The research shows that nurse staffing, including absolute level and licensing as a registered nurse, has an effect on patient outcome.”
Using data from the Datix incident service, the VitalPac patient data system and Allocate workforce software, the study was able to show wards with a higher ratio of registered nurses to healthcare assistants had less slips, trips and falls. Those with a higher HCA establishment had a higher than average amount of falls.
The possibility these findings could by observed by chance was said to be one in 10,000.
The study also showed there were fewer incidents of nausea and vomiting on wards where there was a total establishment of 30 or more full time equivalent nurses.
According to the research, replacing six HCAs with six registered nurses on the six wards with highest incidents of falls could decrease monthly total falls at the trust by 15 per cent.
Following the study, published in by BMJ Open today, UHCW required all patient observations to be done by a registered nurse instead of HCAs due to the strength of evidence this was linked to better outcomes.
In its conclusions the study said: “The relationship between staffing and outcomes appears to exist. It appears to be nonlinear but calculable and a data driven model appears possible. These findings could be used to build an initial mathematical model for acute staffing which could be further tested.”
Professor Leary said the research confirmed the correlations found by other studies of nurse staffing but using patient level data.
She said: “It shows that patterns in this data exist which can help explain the complex relationship between staffing and outcomes, and it helps us understand the nature of the relationship, not just that there is one.
“It’s the next step on from describing associations and a big step to looking at causation.”
Professor Leary said the study relied on good quality patient level data. “Currently not all trusts collect this level of data. We need to think about the data we collect and its use in a far more realistic way. Currently a lot of data is collected and warehoused or data that is not sensitive is collected,” she added.
The research also cast doubt on the usefulness of the new care hours metric proposed by the Carter review as a safe staffing tool. The metric uses an aggregate average to record the hours spent by nursing staff on patient care.
The paper said: “Using a fairly simplistic arithmetical approach to what appears a complex relationship is unlikely to reflect the real world situation or form the basis of a robust model.”
Professor Leary added: “This study and numerous others tells us the relationship between staffing and safety is a complex relationship; it’s hard to see what use an aggregate average will be in real life.
“Using an average at an organisational level is likely to be comparing apples and oranges. Averages are just not sensitive enough.”