Efforts to cut emergency readmissions typically focus on so called “frequent flyers”, but analysis shows screening the wider population may be more effective. Simon Rowe explains

The financial challenge that each primary care trust faces over the next two years is clear – there is a combined need to get more from current services and to reduce the total level of paid activity types. Emergency hospital admissions are an example of the latter.

They will feature heavily in primary care trusts’ plans to save money, yet are historically an area in which the total level of paid activity per financial year has increased. How confident can one be of their contribution to efficiency plans if a reduction in the total has not previously occurred? Do we need to consider viewing emergency admissions in a different way?

Yes we do, according to analysis undertaken by NHS Wakefield District.

PCTs’ plans to cut emergency admissions are likely to focus on the belief that a small number of people account for a disproportionate number of those admissions, and that admissions for long term conditions form a sizeable proportion of the total.

Such beliefs are understandable. There is no shortage of guidance with the reassurance that a focus here will reduce admissions. To an extent this is correct. It would not be hard, for example, to find that admissions for particular individuals or disease areas have reduced by a small amount against the previous year. The difficultly is showing such reductions against a reduction in the total paid number of emergency admissions.

So what do PCTs need if they are to stand a realistic chance of reducing the total number of emergency admissions? The concept of how to prevent emergency admissions is much more intangible than that for which current thought systems and applied commissioning arrangements allow.

The data exercise and NHS Wakefield District

The data from NHS Wakefield District’s work can be used to illustrate a number of points.

Table 1 shows an example in which the total number of emergency admissions is 12. When the data is split by the number of unique individuals who have experienced at least one admission within the studied time period, the total population admitted (in this case, 4) may be compared with the total number of admissions (in this case, 12).

From the number of unique individuals, three have experienced two admissions or more. The second admission may be termed the “first repeat”. Following this same rationale, there are two second repeat admissions (table 2); the remaining three admissions – for the person who had six admissions – may be classed as remaining repeats. The total number of repeat admissions equals the sum of the first, second and remaining repeats. The number of single admissions may also be shown. This is calculated by subtracting the number of first repeat admissions from the total population admitted. For this particular example the single number of admissions is one.

Table 1: Data Extraction

Individuals – pseudonymised identifier (not NHS number)Number of admissions (given financial year)
123451
134672
156783
176526
Total12

 

Table 2: Summary of example data approach

Admission typeNumber of admissions
Total population admitted4
First repeat3
Second repeat2
Remaining repeat3
Total repeat admissions8
Single admissions1
Total admissions12

The combined result of tables 1 and 2 reveals the total number of admissions split between the number of people admitted and the number who experience numerous admissions, by type.

The findings from the exercise for NHS Wakefield District are shown in table 3. It uses data for the PCT’s registered population, and is inclusive of all ages and admissions to all acute hospital trusts. There are two significant findings. The first is that the single admissions for both 2009-10 and 2010-11 total more than half of the total number of emergency admissions; this means most of the total population admitted for the two separate financial years were only admitted once. The second is that the individuals who experienced a second repeat admission (or beyond) form a small minority of the total population admitted in each financial year.

Table 3: NHS Wakefield district emergency admissions by financial year for the registered general practice population

 

Number of emergency admissions

2009-10

Number of emergency admissions

2010-11

Total population admitted26,59526,927
First repeat6,3646,598
Second repeat2,3092,356
Remaining repeat2,4462,527
Total repeat admissions11,11911,481
Single admissions20,23120,329
Total admissions37,71438,408

Data collected during June 2011

These findings highlight two significant points: a focus on frequent flyers, say from the second repeat admission onwards, misses the vast majority of admissions that form the total. In addition, such admissions are unlikely. Basic probability indicates that the chance of an individual having three admissions, for example, is less than one. Admissions occur by exception – that is they will be difficult to prevent.

A further four separate exercises were undertaken and found the following:

The primary clinical reason for up to a sixth of emergency admissions in 2009-10 and 2010-11 was due to a long term condition (from a defined set of long term conditions, based on the codes from the tenth revision of the International Statistical Classification of Diseases and Related Health Problems – it does not include mental health). Further analysis for all admissions for individuals with at least one long term condition admission found a similar result.

  • The remaining five sixths of emergency admissions were evenly spread across a large number of clinical areas/tariff types.
  • Nearly two thirds of the emergency admissions in 2009-10 and 2010-11 for long term conditions involved people who experienced only a single admission in these periods.
  • Of the total population admitted for all emergency admissions in 2010-11, roughly two thirds had no history of emergency admissions. They had no single emergency admission from April 2007 to (and including) March 2010. This can also be shown for 2009-10.
  • Roughly three quarters of all repeat admissions in 2010-11 and 2009-10 were emergency readmissions that occurred within a 30 day period of the previous admission.
  • Between 2008-09 and 2009-10, and 2009-10 and 2010-11 an increase/decrease in emergency admissions for the whole NHS Wakefield District was caused by a matching increase/decrease in the total population admitted and the number of repeats for two thirds of practices. For one fifth, a match was present for the total population admitted only; for the remainder the match was for repeat admissions only.
  • The best chance of reducing admissions at a practice level within a financial year is the same as that for the entire local healthcare economy – to focus on a population size bigger than the expected total population admitted.

Data patterns

The approach taken by NHS Wakefield District has found a number of patterns and rules in the data. Prominent within these is that the number of single admissions always equals the total number of admissions minus the sum of first repeat and repeat admissions. Such patterns point to possible benefits of setting standards for practices in GP clinical commissioning groups and data modelling approaches.

These findings do not represent universal doom and gloom. Steps can be taken to reduce the total number of emergency admissions if enough is known about the patients concerned. NHS Wakefield District’s analysis may help by being applied in other PCT areas.

It would be easy to take one of two standpoints. The first would be to ask “why bother?”. If more than half of emergency admissions are for patients who are only admitted once in a financial year, this must be a “good” system. It may be so – but this is an afforded position, given financial pressures. It is also bound up with an approach that focuses on frequent flyers, with the assumption being that it is “good” if the majority of individuals are only admitted once, but “bad” if a minority are admitted more than this.

The other standpoint is simple: if a lot of people are creating demand in one part of the system – ie being admitted – the capacity of response elsewhere needs to be increased. An example would be increasing the capacity of, and access to, primary care. This is likely to be only part of the story as it assumes a lot about individual behaviours and decision making. Some people simply do not behave in ways we can predict.

There is a need to move away from the current approaches to prevent emergency admissions. The pattern of emergency admissions, within and across financial years, is much more chaotic at the individual level than we thought. It would be difficult, for example, to predict a sizeable proportion of the population who only have a single admission within a financial year.

Applying an approach akin to health screening may be a better idea. Screening works on the premise that a number of people with a given health condition do not know they have it so, to identify them, a larger number must be screened. When translated to admission prevention, this could mean an entire practice population receives scheduled levels of contact asking how they are and whether they need anything in particular from the practice.

This may be seen as a waste of time and money, but can generate information and a level of understanding about a whole population. This is desperately needed to achieve a scale and nature of response that stands a realistic chance of reducing the total number of emergency admissions