Home   About Kurtosis   Courses   Course Calendar   Booking Information   Ideas   Technique   Clients   Contact
     

Five observations about average percentage bed occupancy

The most widely-used measure of hospital bed utilisation can be problematic

One of the most important things we have to do, particularly if we work in a hospital setting, is measure and describe how full hospital beds are. The trouble is, the indicator we tend to reach for in order to do this is average percentage bed occupancy. And average percentage bed occupancy brings problems with it. Here are five observations about those problems.

1. The calculation is itself problematic

Average percentage bed occupancy is a number based on quite a long-winded calculation, which means that, by using this calculation, we are effectively excluding a lot of people from an important debate. This is because average percentage bed occupancy is an abstraction and many NHS managers have difficulty with abstractions. They prefer data when it describes reality in more “direct”, uncomplicated ways. Average percentage bed occupancy doesn’t do that. In case you doubt that statement, let’s have a closer look at how we calculate it.

In order to measure it for —say—General Medicine in April, we take a series of “snapshots” of the number of General Medical patients who were occupying a bed at midnight (midnight!!!) for each of the 30 days in the month. We then add up all those 30 midnight snapshots and that gives us our numerator: occupied bed days. Then we have to calculate our denominator, which is called available bed days, and is basically the bed complement for General Medicine multiplied by 30 (the number of days in April). But we also have to remember to take account of any days that wards or beds might have been closed during April (we’d subtract those from our figure). And we've also got to make a decision about what to do with “borrowed” and “lent” bed days, which in itself can be quite problematic. That gives us our denominator. With our numerator and denominator in place, we can then divide one by the other, multiply by 100 to convert it into a percentage, and that gives us our number: average percentage bed occupancy.

That is a lot of steps for busy people to understand and a lot of places where the calculation can go wrong (analysts are sometimes surprisingly “disconnected” from the bed complement and every hospital has got at least three wards on whose bed complement nobody can ever agree, so it is quite easy for things to go wrong here).

2. The problem with averages

The second observation is that if you quote only this number as the way of describing how you use your beds, then you are assuming—somewhat rashly—that not only do people understand what it means and how it is calculated, but also that they can get their heads around the issues to do with using an average to describe something that might vary a lot. Average bed occupancy of 85% for example can mean a lot of day-to-day variation or it can mean only a small amount of day-today variation. By quoting just the one number you are assuming that people get this. They generally don't.

3. The connected-ness of average bed occupancy

Thirdly, a lot of people don’t realise that average percentage bed occupancy is closely related to other indicators. In fact, one of the beauties of average percentage bed occupancy (for those people who understand it) is that it helps connect together other hospital activity indicators. If you have a 24-bed ward, for example, that admits an average of 20 patients a day at an average length of stay of 1 day, then each day will generate an average of 20 bed days, and if you divide 20 by 24 and then multiply that by 100 you get 83.3% average bed occupancy. Most analysts "get" this. Some clinicians "get" this. But many managers don't. Not because the arithmetic is beyond them but because it relies on them implicitly making the connections between the numbers. Most don’t bother to make the connections.

4. Percentage baggage

The fourth observation to make is that average percentage bed occupancy—like anything measured as a percentage—carries baggage with it. Worryingly, some managers tend to think in over-simplistic terms here: high average percentage bed occupancy is good and low average percentage bed occupancy is somehow wasteful. This is a variation on the joke about the optimist seeing a glass half-full, the pessimist seeing a glass half-empty, and the accountant who sees a glass that’s twice as big as it needs to be. For as long as we use only average percentage bed occupancy to describe bed use, then we are playing into the hands of the folk who take this simplistic view of percentages. This doesn't help us move the debate forward.

5. You need to know what “wrong” looks like

Finally, there is a danger in looking at an average percentage bed occupancy figure if you don’t have any context, if you don’t have a reference–point for whether it’s the “right” figure or the “wrong” figure. If you want to know what the “right” level of bed occupancy is, then you need to know what “wrong” looks like. This may seem like a glaringly obvious statement, but you do have to be clear about what the consequences of the wrong average percentage bed occupancy are. If the wrong level of bed occupancy means cancelled operations, too many patients waiting too long in A&E for a bed to become available, or too many outliers (patients in the wrong beds), then you need to be disciplined enough to define it as such and then to set about doing some analysis and modelling that enables you to find out what occupancy level is consonant with reducing those numbers to acceptable levels. When you’ve done that, you will have created a benchmark that provides you with much-needed context for interpreting your average percentage bed occupancy figures.

So: using average percentage bed occupancy in isolation is fraught with problems. It follows, therefore, that we need to find new, better, more user-friendly ways of describing and visualizing bed occupancy to managers and clinicians. We will, for example, have to probably pick a time other than midnight to measure it. We will need to find a way that makes clear the day-to-day, hour-to-hour variation. We will need to show explicitly how occupancy relates to the other activity indicators and how and when it is associated with dysfunction.

Will there be beds for me and all who seek? A one-day course for analysts.

[28 May 2013]

   
     
© Kurtosis 2013. All Rights Reserved.