emergency admissions: You might think You have heard it all before on rising emergency admissions.But Neil Pettinger argues that once short-stay admissions are taken into account, the figures are not so bad. It is the NHS's recording methods that are wron

One of the most pervasive trends in the NHS over the past two decades has been the unrelenting increase in emergency hospital admissions. This increase has caused problems for the NHS not only because of the sheer number of admissions involved, but also because their unpredictable nature has meant that other hospital services (in particular, elective admissions) have been disrupted as a result. Furthermore, each winter brings the problem into even sharper relief as the volume of admissions increases. And the problem is compounded by the NHS's continuing failure to solve the problem of delayed discharges, since these involve patients who are 'blocking' the very beds needed to accommodate emergency admissions.

This growth in emergency admissions has been well documented, particularly in Scotland where the terrain has been clearly mapped out for some years. By using linked datasets, the familiar stumbling block of double counting episodes has been laid to rest and the dimensions of the problem can be more easily understood.

1In general terms, the rise in emergency hospital admissions has been 2-3 per cent per year over the past two decades, with no real sign of abatement.What the research has also revealed is that the lion's share of the increase has taken place in medical, rather than surgical, specialties. In Scotland, between 1981 and 1998, medical emergencies increased by 95 per cent whereas surgical emergencies increased by a 'mere' 42 per cent.

If the ever-rising tide of emergency admissions has become an all-too-familiar sight on the NHS landscape, then an equally familiar feature must also be the lack of an overarching explanation for it.

While there has been a plethora of partial explanations, there has been no successful synthesis. The end result is a somewhat fragmented 'multi-factorial' explanation that includes such causes as the ageing population, growing patient expectations, enhancements to the capability of medical technology, the shifting pattern of primary care, fear of litigation and changes in household structure. Opinions have varied about the strength of each of these explanatory themes, with the result that no clear way forward to finding solutions to the problem has yet been found.

But one obstacle to developing a more cogent theory may well lie in the terminology. The very way in which data about hospital admissions is classified may be impeding attempts at explaining the phenomenon. It is also possible that one of the health service's traditional descriptive statistics (or performance indicators) also makes finding a solution more difficult.

Taking data classification first, one of the interesting aspects of the way in which the health service categorises hospital admissions is that it treats elective work in a different way to emergency work. Few now regard the dramatic increase in elective hospital admissions over the last 20 years as noteworthy.Yet the increase has been at least as significant as that of the emergency workload. The reason why this increase is never remarked on is that elective workload is, by convention, split into inpatient and day-case work (see chart 1).

During the early 1980s, the 'inpatient' classification began to be reserved for elective admissions that involved one or more overnight stays; short-stay cases were re-packaged as day cases, taking them out of the equation. It is now accepted that day cases are counted and recorded separately from inpatients.

If, however, at the same time as elective work was separated into day cases and inpatients, emergency work had been similarly differentiated, then the health service would by now have got used to looking at trends in emergency activity similar to those in chart 2. In other words, trends not dissimilar to those in elective workload.

It is not straightforward to separate short-stay patients from the rest when it comes to emergency workload.

Notwithstanding debates over the difference between a day case and an outpatient attendance, one of the reasons why the definition of a day case (not in bed overnight) has worked is that it applies reasonably well to the way in which elective workload is carried out in hospitals. Only with the introduction of 23-hour day-case facilities is this classification coming under threat.

With emergencies, however, a 'not in bed overnight' definition is too simplistic. Unlike most elective admissions, emergency admissions take place around the clock, so it would be unwise and unrealistic to exclude patients admitted late in the afternoon or the evening from a short-stay definition simply because they still happened to be occupying a bed 12 hours later. On the other hand, it is impossible to tell from national data sources (and many local data sources) what time of day admissions took place.Hence the definition used in chart 2 is admissions with a length of stay of one day or less. In practice, this means that patients with a length of stay of up to 47 hours will be included.

But it is unlikely that many of the patients will have stayed as long as that.More likely is that the median length of stay will lie between 12 and 36 hours.

There are two salient points to derive from the trend lines: one is that the bulk of the increase in emergency admissions has been accounted for by these short-stay admissions, just as is the case with elective workload. In many hospitals, this will be tantamount to the increase being 'soaked up' by the acute admissions wards.

In theory, therefore, so long as a hospital admissions ward has the right number of beds, and has the right protocols in place for ensuring a steady throughput, the main inpatient wards should be shielded from much of the growth in emergency admissions. But in reality, no such thing has happened. It is possible, however, that a better understanding of extra admissions may help hospitals improve the configuration of their wards.

The second point is that some of the residual increase in longer stay patients (which could arguably be said to reflect 'genuine'morbidity) begins to look as if it might be attributable to an ageing population. In Scotland, the increase of 51 per cent in longer stay admissions between 1981 and 1998 is not a million miles away from the increase of 21 per cent in the population aged over 75 over the same period.

But this realisation that most of the increase has taken place among short-stay admissions has other implications.We still lack an all-embracing theory of why such admissions should be growing at all; all that we have is an assortment of disjointed theories and hypotheses. But an argument could be made that because most of the increase has been in shortstay admissions, this lends more support to the supply-side factors than the demand-side ones. It may now be possible to suggest that it is more likely to be factors such as enhanced technology, changes in the way in which primary care is delivered, fear of litigation and inappropriate referrals that will explain the dramatic rise in short-stay cases. In other words, the phenomenon of increasing medical emergency admissions is perhaps more likely to be explained by changes in the service rather than changes in the population.

Finally, a point about the way in which we look at hospital admissions data. By and large, the length of stay of our inpatient activity tends not to be examined too closely.That doesn't mean that no heed whatsoever is paid to it.Quite the opposite - many senior managers and clinicians are only too aware of where they stand in length-of-stay 'league tables'. Indeed, length of stay has often become a battleground for managers and clinicians, with over-simplistic assumptions being made that short lengths of stay are a good thing and that longer lengths somehow represent outdated clinical practice. But regardless of the rights and wrongs of the debate, length of stay has tended to be described using one statistic only: mean length of stay.

In fact, the familiar trend of falling length of hospital stay over the years has often meant nothing more than an increase in the number of inpatients with very short lengths of stay.The length of stay of other inpatients (whose stay lasts longer than two or three nights) has remained relatively stable.Yet this trend has remained largely invisible because of the way length of stay has been described almost exclusively by reaching for the mean.

The primary message here is that the growth in emergency medical admissions is mainly accounted for by an increase in short-stay admissions. But there is also a secondary point - we need to be more discerning about the statistics that are chosen to describe health service activity.The NHS's obsession with mean length of stay has blinded it to a trend of great importance.

Key points

The rise in emergency admissions has been 2-3 per cent over the last 20 years and shows no sign of abatement.

Medical emergencies account for most of the increase l Emergency admissions should be divided into short and long stays.

This type of analysis shows that the bulk of the increase in emergency admissions has been for short-stay patients.

Better understanding of emergency admissions should help hospitals improve their ward distribution.


1Clinical Outcomes Working Group. Clinical Outcome Indicators. Edinburgh, Scottish Executive, 2000.