Bed occupancy 3: How is it calculated?

Introduction

This is the third segment in a total of four posts about bed occupancy. The first asked why it is important. The second post looked at factors that affect bed occupancy, and the third (this one) provides a bit of information on how it is calculated and reported.

Calculating bed occupancy

All NHS Boards use an electronic record for both tracking their admissions, discharges, and also recording all information about someone's care. In some cases, it will be the same system and in other cases there will be two different systems. For example, NHS Lothian uses a system called 'Trakcare' to record almost all aspects of patient care and it is their main Electronic Patient Record (EPR). NHS Tayside uses Trakcare for patient contacts and admissions/ discharges, whilst all patient notes are recorded using a system called EMISWeb.

The principles are the same, however. When a patient is admitted, that admission is recorded on the  system with an admission date of the date that they are first admitted. At midnight their presence is recorded as an occupied bed day (OBD). When a patient is discharged, that date is their discharge date and their length of stay (LOS) is simply the number of days that they were on the ward (discharge date minus admission date).

Reporting bed occupancy

Since bed occupancy is the no. of admitted patients / no. of beds, expressed as a percentage it should be easy to understand how occupancy reflects demand and activity. However, the no. of beds can vary.

Bed unavailability

If a patient damages a room then that room may be out-of-service for a few days whilst it is repaired since no-one can be admitted to a damaged room. Similarly, renovations such as security improvements are often made to rooms on a rolling basis (e.g. two beds at any given time in a ward).

But unless that room unavailability is updated on the system, it may be counted as an available room. So if you have twenty beds but two of which are unavailable due to damage, you will only have eighteen available beds. If you have eighteen patients, then occupancy is 100%. But the system may think that you have eighteen patients and twenty beds and report occupancy as 18/20 = 95%.

Usually, the no. of admitted patients (i.e. the numerator) is likely to be correct but the only way to get accurate bed numbers (i.e. the denominator) is to get the information from a contemporary source such as daily 'safety huddle' records where unavailable beds are updated.

This is relevant when we try and understand how NHS Boards have provided the data. It is possible that differences in how they count beds affects the numbers. However, as we will see, I have tried to get both the numerator (no. of patients) and denominator (no. of beds) separately and to calculate occupancy rates myself.

Effects of COVID measures

When the first COVID-19 lockdown was announced (March 2020) it became clear that there was a high risk of harm to psychiatric inpatients from catching COVID. Indeed, the risk from COVID was likely to be higher for many patients than the risk of not being admitted. Based on available information at the time, the risk of death from COVID was higher than the risk of suicide. Fortunately, we now have more accurate risk data from COVID, but at the time it was true for many patients (for example, older and overweight men with multiple comorbidities such as diabetes and COPD).

As a result, many NHS Boards made sections of their wards (or whole wards) their designated COVID ward so that all inpatients with COVID could be managed in the same ward; thereby reducing the risk of spread to other patients (and staff).

If you designated a ward of twenty beds as your COVID ward, the beds were not available for general admissions of patients who did not have COVID. However, they may still have been recorded on the system as if they were. So there is a possibility that your occupancy figures will be unrealistically low because if you had five admitted patients with COVID in a ward of twenty beds, your reported occupancy would only be 5/20 = 25%.

These beds were not closed per se but they were not used for other admissions due to clinical decisions based on risk. This is likely to explain the observed drops in occupancy at the start of the COVID lockdown.

Very short admissions

In specific situations such as having someone admitted early in the morning and then discharged before midnight the next day, followed by another admission, you may find that there is more actual work than your occupancy figure suggests. This is because your occupancy (100%) is simply the number of people in a bed at midnight but you will have had two admissions and one discharge whilst another full ward will have the same occupancy (100%) but no admissions or discharges.

Locally, approximately 4% of admissions have a zero day LOS. Therefore, when considering actual 'work', admissions and discharges also need to considered alongside occupancy.

Who does the reporting?

In almost all cases the actual numbers reported to committees, Board meetings, or via Freedom of Information (FOI) requests come from a different department within the NHS Board; not the people who may have the detailed knowledge about day-to-day changes in bed numbers.

Most NHS Boards will have a 'business intelligence' department who can extract the data but it is quite common for those reporting the figures to have little knowledge about how beds are used in practice or who is actually in them.

For example, there are different specialty codes in use across NHS Scotland and if a CAMHS patient (G2) is admitted to a General Adult Psychiatry (G1) ward, the bed will be filled but not by a GAP patient. This can create slight variations in perceived demand: if your GAP ward is 50% of full of CAMHS patients, you may think that there is increased demand for GAP beds but the demand is actually coming from CAMHS.

An occupied bed in a GAP ward does not, therefore, automatically mean that it has a GAP patient in it.

Psychiatry specialty codes

The following is a list of the most commonly-used psychiatry specialty codes.

G1 General Psychiatry (Mental Illness) (GAP)
G2 Child & Adolescent Psychiatry (CAMHS)
G3 Forensic Psychiatry
G4 Psychiatry of Old Age (POA)
G5 Learning Disability (LD)

Reporting the wrong thing

Issues with coding are quite common. Historically, when different NHS Boards compare the number of psychiatric beds they have, they usually group together beds for all specialties even though the specialties very rarely share beds. These figures are usually reported to the Scottish Government (ISD but now Public Health Scotland) so assumptions are carried through the system and errors become lost.

Not all NHS Boards have forensic beds since these facilities are typically commissioned on a regional (or national) basis. This means that those NHS Boards with large numbers of forensic beds (e.g. NHS Tayside) will appear to have more psychiatric beds than other NHS Boards that don't have forensic or LD beds. However, they are only 'hosting' those beds for other NHS Boards - but they will appear on their bed numbers.

NHS Tayside has an inpatient unit for CAMHS but this is commissioned (and funded) regionally. On paper, NHS Tayside has 12 CAMHS beds but only about 5 are paid for by (and belong to) NHS Tayside. The rest belong to other Boards in the North of Scotland catchment area.

Until this is understood, there is a risk of comparing apples with oranges and coming to the wrong conclusions. Unfortunately, it is often misunderstood and NHS Boards will spend lots of time trying to figure out how to reduce their bed numbers because they've been told by those that count beds (wrongly, in many cases) that they have more than other NHS Boards.