Five English hospital trusts have higher than expected deaths, finds new mortality measure

BMJ 2013; 346 doi: (Published 25 January 2013) Cite this as: BMJ 2013;346:f554
  1. Nigel Hawkes
  1. 1London

Five hospital trusts in England have shown higher than expected mortality ratio for two years running, the NHS Health and Social Care Information Centre has reported.1

Colchester, Tameside, Blackpool, Basildon and Thurrock, and East Lancashire registered higher death rates for two consecutive periods of 12 months, 1 July 2010 to 30 June 2011 and 1 July 2011 to 30 June 2012. A further seven trusts had a higher than expected mortality in the most recent period, but attention is likely to focus on the group of five, because two successive years of poor results is more likely than a single year to signal poor care.

Over the same two year period 11 trusts had a lower than expected mortality ratio. They were James Paget in Great Yarmouth; Addenbrooke’s in Cambridge; Kingston; Barnet and Chase Farm; and seven trusts in inner London: the Royal Free, St George’s, the Whittington, Chelsea and Westminster, University College, North West London, and Imperial College.

Hospital mortality is measured by the ratio of actual hospital deaths to the number of expected deaths were the hospital to have the average for England as a whole. The average hospital would score 1, and most others would lie close enough to 1 for no conclusions to be drawn about their quality of care.

The field was pioneered by Brian Jarman, the originator of hospital standardised mortality ratios (HSMRs), which are published by the healthcare information company Dr Foster Intelligence, but the Health and Social Care Information Centre now uses its own measure, the summary hospital level mortality indicator (SHMI). This defines results as being higher than expected or lower than expected if they lie more than two standard deviations above or below the baseline.

The SHMI counts all deaths in hospital or within 30 days of discharge, and the expected number of deaths is corrected for factors such as case mix, age, source of admission, and comorbidities, so that hospitals with unusually ill patients are not disadvantaged. But the SHMI does not correct for palliative care, which has in the past distorted HSMR measurements. The information centre said simply that a greater proportion of patients had been recorded as undergoing palliative care in both the higher than expected and lower than expected groups than in trusts generally, without quantifying what this meant. It also pointed out that hospitals that did poorly had more patients from deprived areas, while those that did well had fewer.

The SHMIs for the year from July 2011 to June 2012 ranges from a high of 1.25 for Blackpool Teaching Hospitals NHS Trust to a low of 0.71 at Whittington Hospital and at University College Hospital, London.

Tim Straughan, chief executive of the information centre, said, “Today’s report, based on two years of data, shows an emerging picture of which trusts are categorised over time as having higher or lower mortality ratios than expected—and indeed also shows that the vast majority of trusts in England have a mortality ratio that is as expected, based on the characteristics of the patients they will typically treat.

“As always with such a complex area, this mortality indicator should be seen as an early warning mechanism, rather than a definitive judgment, to examine the reasons why a trust’s ratio is higher or lower than expected.”


Cite this as: BMJ 2013;346:f554