New academic paper: Stafford HSMRs NO indicator of avoidable deaths

Please share this widely. It’s absolutely crucial for the future of Stafford hospital and of the wider NHS.

An important new academic paper was published this week by the Oxford University Press‘ Oxford Journals on the subject of Hospital Standardised Mortality Ratios (HSMRs). This paper, by a team of respected academics from the ‘Primary Care Clinical Sciences’ and ‘Public Health, Epidemiology and Biostatistics’ departments of the University of Birmingham is titled:

The findings of the Mid-Staffordshire Inquiry do not uphold the use of hospital standardized mortality ratios as a screening test for ‘bad’ hospitals

This paper examines the findings of the Francis Report into events at Mid Staffs NHS Foundation Trust – and reaches two findings that are absolutely vital for the proper public understanding. The whole paper can be read here, but the key conclusions – quoted from the report with the permission of its principal author – are as follows:

Targeting hospitals for high HSMRs is invalid

The report’s summary states that

The initial investigations were triggered by an elevated hospital standardized mortality ratio (HSMR). This shows that the HSMR is being used as a screening test for substandard care; whereby hospitals that fail the test are scrutinized, whilst those that pass the test are not.

As I’ve written on this blog, 14 hospitals (10% of all hospitals in England) are already under special investigation because of supposed similarities to Stafford in their HSMR statistics – even though, by the more reliable SHMI measure, 9 of them do not have high mortality rates. Not only this, but a total of 36% of the country’s hospitals are set for similar investigations for the same reason.

As I have argued, so this academic study has also concluded: HSMRs do not indicate avoidable deaths and cannot validly be used for the purpose of identifying hospitals with a ‘needless death’ problem.

Mid Staffs death claims are spurious and damaging

Here is the report’s conclusion, reproduced in full:

The widespread use of HSMRs as a screening test for clinically avoidable mortality and thereby substandard care, although well intentioned, is seriously flawed. The findings of the MSFT inquiry have no bearing on this conclusion because a ‘bad’ hospital cannot uphold a bad screening test. Nonetheless HSMRs continue to pose a very sombre public challenge to hospitals regarding the quality of their care, whilst the unsatisfactory nature of the HSMR remains a largely unacknowledged and unchallenged private affair. This asymmetric relationship is inappropriate, unhelpful, costly and potentially harmful. Using the HSMR to identify ‘good/bad’ hospitals is analogous to the practice of dowsing—the search for water without scientific apparatus—it is time to abandon this screening test and search for a better one. Meanwhile, the use of process measures remains a valid way to measure quality of care.

This is an explosive conclusion – that the use of HSMRs to claim ‘needless deaths’ is as inaccurate, invalid and unreliable as searching for water using a twig. Not only that, but any such use has the potential to be damaging.

These findings mirror my own articles (and are put not much less emphatically!). There is growing consensus among statisticians, academics and, increasingly, among sections of the media (the Straight Statistics and Computer Weekly websites and even, very quietly and cautiously, one BBC News writer) that the claims of ‘excess mortality’ at Stafford hospital are – at best – utterly spurious and inaccurate.

And yet they continue to be recited, rote-fashion, as simple, unchallenged fact by the right-wing press and even by the BBC (just this evening on Look North during a segment on the inspectors’ visit to North Cumbria Hospitals Trust, one of the 9 hospitals targeted in spite of having average or lower SHMI mortality rates).

The weaknesses of HSMRs and of the data entered to create them are not difficult to identify. The articles on these weaknesses are not encrypted or hidden. And yet they continue to be conspicuous by their absence in the mainstream media while false figures are recited as fact.

You really do have to wonder just why that is.