Bed occupancy 4: General adult psychiatry 2018-2023

Method

A freedom of information (FOI) request was submitted to all mainland NHS Boards in February 2023. Only mainland Boards were selected since Island Boards typically admit patients to the mainland, and they will not have their own beds.

The text of the request was:

For each calendar month (starting on 1 January 2018 and ending on 31 January 2023), please provide:

1. The total number of admission beds available for 'General Adult Psychiatry' which are grouped as 'acute' psychiatric wards. Under the ISD Data Dictionary code, please include only code 'G1'. Please *exclude* all beds that are allocated explicitly to: rehabilitation; substance misuse; forensic (G3); psychiatry of old age (G4); learning disability (G5); and child/ child and adolescent psychiatry (G2, G21, and G22). If possible, please group by Ward for each month.

2. The total number of occupied beds for each ward by calendar month. This will typically be the occupied beds at the midnight 'census' using whatever bed management system is in operation. Please group by ward of admission and not the speciality code for that admission (so LD admissions to a general psychiatry bed are still counted).

3. The total number of bed days in each ward that are occupied by delayed discharges, irrespective of reason/ code.

Only GAP (G1) beds were requested as I am a General Adult Psychiatrist and was mainly interested in GAP.

I wanted to be able to calculate the no. of beds per capita for each NHS Board, and I could calculate this if I knew the total number of available bed days in any given month.

Bed occupancy can be calculated easily from the data I requested, and asking for the raw data gave me more flexibility in calculating additional information.

Finally, I asked for information on delayed discharges as I was interested in the total proportion of bed capacity that was lost to delayed discharges.

Results

Description of the data

All NHS Boards provided the data requested, with most providing it within the statutory timeframe. One (NHS Ayrshire & Arran) emailed before the deadline to apologise for the delay and they provided it a few days later.

Not all NHS Boards could provide all the data for the whole period requested. NHS Forth Valley did not have available bed data for the period Jan '18 to Apr '19, although they did provide data on occupied bed days (OBDs) for this period. Reasons were attributed to changes in electronic record systems during the reporting period.

Data on delayed discharges was only provided by eight NHS Boards (NHS Tayside, NHS Dumfries & Galloway, NHS Forth Valley, NHS Borders, NHS Lothian, NHS Grampian, NHS Lanarkshire, and NHS Ayrshire & Arran). In two cases data were missing for early parts of the time period: NHS Tayside (Jan '18 - Feb '19); NHS Lanarkshire (Jan '18 - Apr '19).

Some NHS Boards provided data for wards that were requested to be excluded. For example, NHS Dumfries & Galloway provided data for their POA wards as well as GAP wards. This was identified from the corresponding beds per population being abnormally high. The actual GAP wards were identified using information from the Board's website.

NHS Forth Valley provided bed numbers (and OBDs) for many of their long-stay/ rehabilitation wards, even though this was asked to be excluded. Again, the actual GAP admission wards (including IPCU) were identified from information available on the Board's website.

In the tables below, NHS Boards are listed in the order that the data were received. The data have been aggregated in MS Excel and conditional formatting has been used to create heat maps, with better figures being represented by green values, and worse figures being represented by red.

Beds per capita

There is usually endless discussion within NHS Boards about whether they have more or fewer beds than other NHS Boards. I have yet to see a credible set of figures from the Scottish Government. My experience is that people argue this from positions of relative ignorance since they do not understand the data they are looking at.

Therefore, I calculated the beds per 1,000 population using the following data:

1. No. of beds available in a given month = total available bed days / no. of days in that month.
2. Population size = Mid-Year Population Estimates, Scotland, mid-2021 from National Records Scotland.

GAP beds per 1,000 population. NHS Scotland (aggregated figures) are in dark blue and the Scottish average is the orange horizontal line.

Bed occupancy

The following table shows monthly bed occupancy by NHS Board. The occupancy for NHS Scotland has been calculated by adding up all the occupied bed days and dividing that by the number of available beds.

Bed occupancy rates for each NHS Board over time

Differences between NHS Boards

The first thing to note is that some NHS Boards have had high occupancy for a long time, whilst others have just got busier in the last year or so.

NHS Tayside, for example, has had high occupancy (the highest average of any NHS Board) since 2018; regularly getting into the high 90's and often exceeding 100%. Only NHS Lothian, NHS Greater Glasgow & Clyde, and NHS Ayrshire & Arran have broadly similar patterns. As noted below, NHS Fife has got busier towards the end of 2022 but was previously running at occupancies in the 80's to 90's.

Some NHS Boards (NHS Forth Valley, NHS Borders, and NHS Lanarkshire) have regularly operated within the 85% upper limit.

These figures have been presented in a line chart below.

Bed occupancy rates for NHS Boards over time. NHS Scotland (aggregated) is shown as thick black line.

Effects of COVID-19

The impact of COVID-19 (discussed in previous posts) can be seen clearly. There was a plateau following the first COVID lockdown (Mar '20) from mid-2020 until the end of 2020. The other observation is that bed occupancy has been clearly rising for Scotland as a whole since early 2021 and in the last six months or so it has been close to 100%.

There are differences between NHS Boards, however. NHS Tayside had a reduction in bed occupancy after the first lockdown in March 2020, due to the segregation of one ward for COVID cases only. Since there never was a massive peak in COVID cases needing psychiatric admission, the ward ran at low occupancy.

All NHS Boards appear to have had a reduction in occupancy after lockdown, although the duration of reduced occupancy varies between Boards. In NHS Tayside, occupancy was back above 85% within six months of lockdown. It lasted a little longer in NHS Fife, and reduced occupancy lasted at least a year in NHS Forth Valley and NHS Lanarkshire. Some of the largest NHS Boards (NHS Lothian and NHS Greater Glasgow & Clyde) saw reductions below 85% lasting only three months or so.

Possible artefacts

There are some anomalies that have not been explained but may well represent data 'artefacts':

1. Why NHS Fife had occupancy rates of sub-90% until Jul '22 when they went above 100% quickly.
2. Why NHS Ayrshire & Arran had a very high occupancy in Aug '21. Their submission reports that for four months around this time they were using a 'Decant Ward (Fireworks)' and whilst some of the other wards had lower figures during this time (suggesting patients being treated in a different, temporary ward), in Aug '21 there was a high total for all the wards. This may be because admissions to the Decant Ward were incorrectly included in more than one ward.

Delayed discharges (DDs)

The proportion of total bed capacity lost to delayed discharges, by NHS Board, is shown below. Again, figures have been aggregated to provide a figure for NHS Scotland MH services as a whole.

Two aggregate figures have been provided: 1) the reported DDs as a proportion of all NHS Scotland beds; 2) the reported DDs as a proportion of only those Boards where DD data was provided with bed data.

Proportion of bed capacity lost to delayed discharges

The total bed capacity of NHS Scotland used by delayed discharges that is recorded is 6.6%. If we assume that the missing figures are similar, and we look across only those NHS Boards that provided both DD information and bed information, we can see that 10.7% of all beds (about 1-in-10) are unavailable due to delayed discharges.

There is, however, wide variation between NHS Boards with some NHS Boards (NHS Dumfries & Galloway, NHS Ayrshire & Arran) having very high figures and others (such as NHS Borders and NHS Grampian) having no delayed discharges for much of the time. Both sets of figures seem improbable.

I think that this variation is best explained by unreliable data: possibly because DDs are either not being recorded consistently or because information about DDs have not been included in this FOI request.

There are anecdotes about delayed discharges being obscured within routinely-reported data. One way to do this is to remove the flag as a delayed discharge from the online system. Since staff will flag an admission as a delayed discharge, others can remove the flag, so these data may be unreliable. Those who want to flag delayed discharges (such as inpatient staff) may have a conflicting set of interests to others who may be responsible for finding accommodation, nursing homes, or other resources needed to allow a patient to be discharged.

It is important to remember that although Integration Joint Boards (IJBs) are responsible for commissioning services, the Local Authority and NHS Board are responsible for providing relevant services and they almost always have separate budgets. It is therefore highly probable that there is tension between the financial needs of different services. Although some (such as the current Scottish Government) may be proposing a National Care Service as a solution, there is no evidence to suggest that this will work and most published evidence indicates that there hasn't been obvious benefit from integration of health and social care (Alderwick, 2021; Kadu, 2019; National Audit Office 2017; Reed, 2021; Rocks, 2020).

Indeed, the figures shown above regarding delayed discharges are very likely to be higher (rather than lower) than can be evidenced above.

Has Scotland really run out of beds?

This story started with a colleague flagging a complete lack of general adult psychiatry beds in Scotland on one particular day in February 2023. I therefore wanted to try and determine how likely it was that Scotland regularly ran out of beds.

In order to work out on how many days there were no beds you would need daily data for each NHS Board. However, it was possible to calculate how many 'spare' bed days were available for each month for each NHS Board. This was simply the total number of bed days - the total number of occupied bed days. You could calculate the average number of 'spare' beds each day by dividing the monthly figure by the no. of days in that month.

These figures are shown below. It is assumed that you can't have negative beds so if an NHS Board had more OBDs than beds, it was allocated 0 'spare' beds that month. The figures were aggregated for the whole of NHS Scotland and this value was divided by the days in the month to get an average no. of 'spare' beds per day for that month. There may be days when there are no beds available, but the average will give you an indication of how 'tight' the bed availability is.

Spare beds per month for each NHS Board

The NHS Scotland total is on the far right. It can be seen that in all months, there is a greater-than-zero figure for the average no. of beds available on each day.

As stated above, it is possible that on some days there are no beds. It is also possible that the figures provided by each NHS Board carry enough uncertainty to hide a complete lack of beds. Some admissions may be misattributed to a different specialty, or there are beds unavailable due to repair or other reasons.

What is more possible, however, is that there are many days when an NHS Board has very few beds and they are unable to provide a bed for another NHS Board. Other NHS Boards are frequently out of beds and they may be told that are no beds available to them.

But... General Psychiatry (G1) includes IPCU wards (Intensive Psychiatric Care Unit) and these should, ideally, run at about 60% occupancy. Most large NHS Boards will have at least one IPCU ward and assuming that these are running at optimum occupancy, there will always be beds in the IPCU. However, these are not separated in the figures above so some of the apparent capacity may actually be in ICPUs in Scotland, and these beds would/ should not be used for general psychiatric admissions; especially not for admissions from other NHS Boards.[1]

Conclusions and reflections

I think it makes sense to add some more detailed discussion about what these figures suggest in a later blog so I will keep the conclusions brief for now.

1. There is wide variation between NHS Boards with regards to bed occupancy.
2. There is still recognisable unreliability to the data which may mean that no-one will ever know what the true state of affairs is.
3. National figures are, by nature, unreliable and I would not advise anyone to believe what they are told without understanding exactly how the data are obtained. As Lenin and/ or Reagan said: "Trust, but verify."
4. Whilst it cannot be concluded that there are situations when there are no beds at all in Scotland, the data would suggest that there are probably some beds somewhere. However, NHS Boards who have beds may be unwilling to give these up to other NHS Boards who may seem unable to return them quickly and consistently.

Notes

[1] Most NHS Boards will probably use up their ICPU beds for local admissions in emergencies - such as when they have no other beds anywhere else. However, when you never have beds there is a risk that IPCU is used routinely as 'overspill' for general psychiatric admissions.

References

Alderwick, H., Hutchings, A., Briggs, A., et al (2021) The impacts of collaboration between local health care and non-health care organizations and factors shaping how they work: a systematic review of reviews. BMC Public Health, 21, 753. https://doi.org/10.1186/s12889-021-10630-1

Kadu, M., Ehrenberg, N., Stein, V., et al (2019) Methodological Quality of Economic Evaluations in Integrated Care: Evidence from a Systematic Review. International Journal of Integrated Care, 19, 17. https://doi.org/10.5334/ijic.4675

National Audit Office (2017) Health and social care integration. London: National Audit Office. https://www.nao.org.uk/report/health-and-social-care-integration/

Reed, S., Oung, C., Davies, J., et al (2021) Integrating health and social care: A comparison of policy and progress across the four countries of the UK. London: Nuffield Trust. https://www.nuffieldtrust.org.uk/files/2021-12/integrated-care-web.pdf

Rocks, S., Berntson, D., Gil-Salmerón, A., et al (2020) Cost and effects of integrated care: a systematic literature review and meta-analysis. European Journal of Health Economics, 21, 1211-1221. https://doi.org/10.1007/s10198-020-01217-5

[Last updated: Thu 6 Apr 10:41]