16 October 2013
Nick Barber, Director of Research, The Health Foundation - photo

Nick Barber

Director of Research
Health Foundation

The focus on the crises in Accident and Emergency has been on inappropriate attendance, staffing and availability of services. However we now have new insights from QualityWatch – the collaboration between the Nuffield Trust and the Health Foundation, dedicated to providing independent indicators of health care quality.

The QualityWatch team has produced a 'Focus on' report on trends in a different area – the conditions that should have been treatable in primary care and should not have resulted in a hospital inpatient admission.

Ambulatory care sensitive (ACS) conditions, as they are slightly obtusely named, are increasingly used world wide as markers of care quality and are now in the NHS Outcomes Framework. We wanted to know what had happened to them in England over the last dozen years.

The QualityWatch team studied over 165 million consultant episodes for ACS conditions from 2001 to 2013 and found stark results; they accounted for one in five emergency admissions. What's more, the number of admissions increased by 48% over the 12 years of the study – more than the growth in other emergency admissions (34%). Why is this?

The growth has been particularly prevalent in the areas that affect the elderly and the young, with half of all admissions in five conditions: urinary tract infection (UTI)/pyelonephritis; pneumonia; chronic obstructive pulmonary disease (COPD); convulsions/epilepsy; and ear, nose and throat (ENT) infections in the young.

Why do you think there is geographical variation?

But the story is not all bad. For example there have been reductions in admissions over time in some conditions, such as angina and bleeding ulcers, and while most local areas showed increased rates of ACS admissions, a small number had made significant reductions.

There are two-fold geographical variations, even when the data is standardised for age, sex and deprivation.

The indirectly standardised admission ratio of observed-to-expected ACS across local authorities varies from 0.65 to around 1.34 – one is, by definition, the national average, and values greater than one have more ACS admissions.

The authorities with the worst ratios cluster most in the North West. When the distribution is shown as a normal distribution, 14 local authorities are more than two standard deviations above the mean – there is around a one in 20 chance this would happen by chance.

So what causes this variation? Chance, underlying health (there is some weak correlation), some oddity of the measure? Or is it quality of care?

We need to do further research to clarify this, however for now the assumption must be that it is quality of care – it seems morally wrong not to at least question local practice if it's shown to be an outlier.

This problem raises an issue that will continue with QualityWatch, and with any other series of indicators or metrics: how should they be interpreted?

Like any measure, they can only be judged fit for purpose by being used and by attending to feedback. Measures need to be honed as much as any other tool and you can help us in this. We will be producing more 'Focus on' reports in the future, and we hope you will help by giving feedback.

You can start now: have a look at the report, in particular page 14. Why do you think there is variation? Are you in an authority with a high ratio? Let us know why you think it is so high. Together, we can improve the system.


Thanks for this report. It has stimulated some questions for this project: I'd be careful about calling ACS admissions 'preventable' - implying the aim should be zero, or that these patients shouldn't have been admitted. Surely some admissions are inevitable even with best ambulatory care. What is the best practice benchmark? Do you want these to be used locally? If so we need to have the local data available (the maps are too small and don't give the level of detail e.g. which conditions in which local area). Do you need to draw distinction between A&E admissions and emergency admissions (including emergency referrals from GP)? Some discussion of planned admissions that could have been avoided could also be useful - e.g. acute deterioration of poorly managed arthritis could turn up as a elective hip replacement rather than A&E admission but still avoidable cost (I think!). Hope that's helpful.
Joe Farrington-... (not verified)
(changed )

Thanks for the comment Joe - you raise many interesting points. I hope we're quite upfront that the report is about ambulatory care sensitive (ACS) emergency admissions - a subset of all admissions that might be considered "preventable". As you point out, the definition of "preventable" admissions can include all sorts of events that are preventable by all sorts of different mechanisms. In the case of ACS, it's important that we don't see the admission in isolation, but consider the care pathway that led up to it. This is where the prevention - or lack thereof - happens. The negative outcome is not that a patient in distress was admitted to hospital, but that their health had deteriorated to a level at which that became necessary. Also for that reason I don't think it's key to separate admissions through A&E from direct GP admission, although that may provide interesting further analysis. I think it's a fair question to ask whether zero ACS admissions is a realistic aim, and the literature tends to expect there will always be some residual level of unavoidable ACS admission. However, that poses the question of that might an "acceptable" level be? As far as we're aware no one has yet attempted to define this yet, but given the amount of variation we observed between areas in ACS admission rates (even after adjusting for the age, sex and deprivation of the local population), it's clear that most areas in England do have avoidable ACS admissions. We're aiming to make the data used in our report available online in the coming weeks, as well as a full list of the diagnostic codes we used so that people can apply our analysis locally on an on-going basis.
Ian Blunt (repo... (not verified)
(changed )

Thanks Ian. I don't think anyone would claim anywhere was already at an optimal (or even acceptable) level of ambulatory care. There is sometimes a simplistic assumption that all ACS admissions are a failure of care - but we should definitely be aiming to bring them down. The issue of ambulatory care sensitive elective admissions may also be worthy of further analysis. I'm sure local data would be useful to commissioners and providers alike as we begin planning for 2014.
Joe Farrington-... (not verified)
(changed )

Add new comment