A week after a healthcare watchdog identified staffing shortages as a key reason for failings at an acute trust, Nursing Times has obtained new evidence confirming a direct correlation between patient outcomes and nursing workforce numbers. Steve Ford investigates
Last week the Healthcare Commission identified chronic staff shortages at Mid Staffordshire NHS Foundation Trust in a highly critical report on the trust’s performance. The report estimated that as many as 400 more people than would have been expected may have died in the trust’s care between April 2005 and March 2008.
There has been debate about the actual impact of nursing staff levels in the NHS for at least a decade but previous studies have always been open to criticism that they may not show the whole picture.
Nursing staff levels have also been identified among the candidates for the new nursing performance indicators, or metrics – called for by the NHS Next Stage Review and currently being developed for roll-out across the health service. However, a subsequent key report, State of the Art Metrics for Nursing, stopped short of recommending nurse numbers as a performance indicator ‘front-runner’, tending to favour more obvious clinical measures of quality such as falls and pressure ulcers.
Now the latest statistical research by Dr Foster, revealed for the first time by Nursing Times, appears to provide the answers missing from previous studies and concludes that the more nurses an acute trust has per bed really does improve mortality rates and length of hospital stay.
The research was carried out by Dr Simon Jones, chief statistician at Dr Foster Intelligence and senior research fellow at the National Nursing Research Unit at King’s College, London, and Professor Peter Griffiths, director of the unit.
They set out to examine what factors influenced outcomes in English hospitals, with specific reference to staffing levels – looking at a total of 147 acute trusts with A&E departments. They compared national hospital activity data for day case and inpatients in England between 1 April 2007 and 31 March 2008 with the latest workforce data from the NHS Information Centre, published last week.
According to this data, of the 49 trusts with the lowest nurse per bed ratios, 43% are rated as having a high hospital standardised mortality ratio (HSMR) – that is to say a patient mortality rate that is higher than expected, based on a typical English hospital. Additionally, among the 49 trusts with a high nurse per bed ratio, 43% had a low HSMR.
Analysis suggests trusts with a low nurse per bed ratio were roughly twice as likely as those with a high nurse to bed ratio to have a high HSMR rating. This is illustrated by the larger number of ‘green’ trusts in the bottom half of the table and the larger number of ‘red’ trusts towards the top.
As might be expected Mid Staffordshire is ranked among the 30 worst performers on nurse per bed ratios. Interestingly in 2007, before the trust had started to react to concerns from Healthcare Commission investigators, Mid Staffordshire would have been ranked as poorly as eighth worst for nurse per bed ratios, using the same statistical model.
Dr Jones explains that his model ‘indicates general trends’ and that there are other factors that also need to be taken into account when interpreting the data. He noted that trusts with low staff to bed ratios did not automatically have high HSMRs. For example, Northumbria Healthcare Trust had the second lowest staff to bed ratio in the country but actually had a very good HSMR. ‘You need to know the number of GPs per 1,000 of the population, the number of qualified staff per bed, and if it’s in London,’ he said.
‘To some extent you need to know all of that in order to conclude why it’s a bad performer,’ he added. ‘You could imagine somewhere that’s a very low nursing staff ratio that might have a phenomenally high GP per 1,000 population, and so the two counteract each other.’
It is, therefore, low nurse levels combined with poor counter-balancing factors, such as a low concentration of GPs locally, that might be an early warning of staff shortages linked to patient mortality rates, that warrants investigation.
However, while other factors can have an impact, the model shows that by far the strongest indicator of trust performance is nursing and qualified staffing levels.
Dr Jones said that the difference in staffing ‘explains about 36% of variation in trust performance on the HSMR’. ‘The rest is explained by other factors, for example, having more nurses is not actually enough – they also actually have to work effectively,’ he said.
Additionally, the model shows that it is not just mortality rates that are affected by staffing levels but also the number of ‘very long stay patients’ – those whose length of stay in hospital exceeds the length of stay of 75% of patients in England with a similar diagnosis and admission type in a certain year. Very long stays have been previously identified as an indicator of ‘deviations from normal care’.
The model suggests that in trusts with 1.4 or fewer nurses per bed, 25% of patients are likely to be very long-stayers, compared with less than 23% in trusts with 1.5 or more nurses per bed.
Although the difference may look small, the relationship is statistically strong, according to Dr Jones. ‘As the number of nurses goes up, the percentage of long-stayers goes down,’ he said.
This new work is significant for a number of reasons. US research has previously reported that nursing levels have a significant impact on mortality but analysts have pointed out that the large differences in staff skill-mix between the US healthcare system and the NHS have made it hard to transfer the conclusions to the UK with confidence.
Previous UK studies have attempted to identify the factors that most likely explain differences in outcomes between NHS hospitals – of which two are probably the significant and best known.
Back in 1999, Professor Sir Brian Jarman, from the Dr Foster Unit at Imperial College London, looked at hospital death rates over four years. He and his team concluded that, rather than the number of nurses, it was the number of medical staff that was most significant.
‘The ratios of doctors to head of population served, both in hospital and in general practice, seem to be critical determinants of standardised hospital death rates; the higher these ratios the lower the death rates in both cases,’ they wrote in the BMJ.
A second study, led by Professor Anne Marie Rafferty, head of the Florence Nightingale School of Nursing and Midwifery at King’s College compared outcomes in general, orthopaedic and vascular surgery with nursing levels at 30 trusts. Published in 2007 in the International Journal of Nursing Studies, it found a ‘large and consistent effect of nurse staffing on mortality outcomes in surgical patients’. However it did not include medical staff and so did not offer comparison of effect between professions.
This latest study aimed to closely follow the structure of Professor Jarman’s original study for ease of comparison of results over time.
It takes into account the many changes that have affected the NHS since Professor Jarman’s work in 1999, including the introduction of enhanced roles for nurses it offers some comparison between the effect of nursing.
Perhaps the key finding was that the most significant factor affecting a hospital’s HSMR was the number of ‘qualified’ staff – either doctors or nurses – per bed.
‘We found that the number of qualified staff, rather than the number of doctors is the best predictor of a hospital’s mortality,’ the authors said. Much of the change could be explained by modernising medical careers and enhanced roles for nurses, they suggest.
Could this then be the first proof that trusts should think of employing more nurses in enhanced roles, and possibly less medical staff? ‘It’s interesting to note that ratio of nurses to doctors was not found to be a significant factor. One possible interpretation of this fact is that trusts can make a limited substitution of doctors by nurses without reducing their HSMR,’ they said. ‘However this interpretation should be treated with caution,’ they added.
Dr Jones emphasises that the model can be further improved and there are many other aspects of hospital care that need to be analysed. He points out that the work is ‘ongoing’.