Mortality rates: getting the right measure

The Secretary of State has recently announced an annual review of the case notes of 2,000 people who have died in hospital every year. Chris Sherlaw-Johnson examines the problems with this approach.

Blog post

Published: 13/02/2015

The Secretary of State has recently announced an annual review of the case notes of 2,000 people who have died in hospital every year. The purpose of this is to identify how many deaths could have been avoided through better care quality – and presumably find ways to identify and implement change where care falls below standard. 

Although we hope that hospital care improves our chances of survival, we also know that simply counting deaths is misleading. For me there are three underlying issues:

  1. Why not use the hospital mortality measures that are used at the moment?
  2. Could you achieve useful measures from just 2,000 case note reviews in a year?
  3. Is it sufficient to focus on mortality alone?

1. Why do we need something more than existing hospital-wide mortality measures?

The existing hospital-wide mortality measures, the Hospital Standardised Mortality Ratio (HSMR) and Summary Hospital Mortality Indicator (SHMI) are derived from administrative hospital records rather than case notes. They are the result of a complex analysis that takes account of individual patient risk so that the results do not adversely affect hospitals when they treat high risk patients. However, they are not, nor have they ever been claimed to be precise measures of quality of care. Typically, a hospital will monitor these alongside a suite of several other measures for the purpose of identifying where there may be problems. They do not measure avoidable deaths, but include all or most deaths that take place in a hospital. However, in a study published in BMJ Quality and Safety, the authors estimated that around 5 per cent of hospital deaths are avoidable. So, although they are not designed to be precise measures of quality, this still raises questions about the extent to which the HSMR and SHMI alone are able detect variations between hospitals.

At face value, therefore, looking for avoidable mortality using case notes seems like a good idea. It not only may provide a more understandable measure, but leads straight to the information (i.e. the case note) that should, in principle, enable people to make judgements about quality of care. Since triggers from the HSMR or SHMI, or any other quality indicator should be leading to case note reviews anyway, why not start with the case notes and build statistics from these?

 2. Could you achieve useful measures from just 2,000 case note reviews in a year?

But case note review takes a lot more time and effort rather than using existing data, and there is the question of whether 2,000 reviews are enough. 2,000 reviews translates into around 13 per hospital trust, which is hardly sufficient to generate comparative mortality statistics – and if only 5 per cent are avoidable there is a reasonable chance that a hospital will find none at all. The Department of Health mention that, at a local level, numbers would be ‘estimated’ rather than calculated. But Professor Nick Black from the London School of Hygiene and Tropical Medicine, whose work has led towards this focus on avoidable mortality, doesn't think this will work. And if the aim is to produce something that is less controversial and easier to understand than the HSMR or SHMI, there is a high chance of not achieving it.

3. Is it sufficient to focus on mortality alone?

Even if a larger number of reviews were undertaken to achieve something more robust, there is the broader issue as to the importance given to an avoidable mortality measure in contrast to other measures of quality. For most people poor care does not result in their death: for example, they may acquire a wound infection, fail to receive vital medication, develop pressure sores, or end up in hospital longer than planned. So, of all the care provided by an organisation, how much would the quality be reflected by numbers of avoidable deaths? We don’t know the answer, and, until we do, we need to be careful about how much emphasis is given to this measure. 

A way forward

So what might be a sensible way forward? It is good to highlight avoidable deaths as a measure for hospitals to monitor with a view to improving care. But to work, case note reviews need to be supported by a robust process for learning, setting actions and monitoring the progress and value of any changes that may result. And it would be most useful if the focus is not exclusively on mortality. However, the reviews should be within the resource constraints of an organisation, and if they could be incorporated into existing internal audit programmes, so much the better. As for comparing hospitals, I believe we are a long way from anything that is likely to work and, most importantly, there is a risk of distracting hospitals from what should be the primary purpose, namely of improving care. 

Suggested citation

Sherlaw-Johnson C (2015) ‘Mortality rates: getting the right measure’. Nuffield Trust comment, 2 February 2015.