How to get better value for money from psychiatric care units
High cost, low volume and long admissions: Julian Walker and colleagues explore the challenge of evaluating treatment costs for patients in medium secure psychiatric units.
Medium secure units accommodate a small proportion of psychiatric patients but consume 1 per cent of the entire NHS budget. The complexity and variety of patients’ needs and length of stay make predicting associated costs difficult. The challenge in the current economic climate is to measure accurately what medium secure units do, compare costs sensibly across service providers and develop benchmarks for quality and value while managing perverse incentives.
Medium secure units are secure psychiatric institutions that accommodate people who have usually committed serious crimes and have severe psychiatric conditions (most commonly psychosis). Their mental illness is often is too great for prison but they are too dangerous for general psychiatric wards.
There are roughly 70 units in the UK and about 5,000 patients, often costing more than £200,000 a year, resulting in annual national spending of some £1.2bn. The number of beds in medium secure units has increased significantly over the last 20 years, partly due to rising demand and the drive to reduce high secure hospital places which are even more expensive.
A quality peer review network managed by the Royal College of Psychiatrists rates the quality of member units by evaluating their compliance to a range of nationally agreed standards on a yearly basis. This review is qualitative in nature, carried out by clinicians and patients from other medium secure units, and incorporates a wide ranging review of policies, procedures, systems, and staff and patients’ views.
However, these reviews do not cover important measureable information such as cost, the number of people assessed, admitted, length of stay or achievement of targets and key performance indicators. This information is key not only for service providers but also for commissioners. Changes in NHS commissioning mean that in 2012 specialist commissioning for forensic services – including medium secure units – will be done by the NHS Commissioning Board rather than regional specialist commissioning groups.
Medium secure units deal with a relatively small number of people at the extreme end of the mental health population and may look after them for years. Such a high cost, low volume service makes looking at typical or average metrics almost impossible. Metrics of high cost, low volume services are also vulnerable to the effects of outliers. These are expensive to treat (for example £2m for a 10-year admission) so a calculation error could have a significant impact on commissioning assumptions and projections.
To compare costs between units, commissioners use the occupied bed day cost as a simple comparator that can be applied to any unit regardless of size, patient type, procedures or any other aspect of the service. The national reference occupied bed day cost for medium security is £451 per day. The problem with bed day cost is that it tells you nothing about how much a patient’s care package costs – for a one month admission the bed day cost would be £13,500; for 10 years, more than £1.6m. Not only is it misleading in terms of an organisation’s cost efficiency but it can also be so if taken as a proxy for treatment episode cost. Inpatient treatment costs need to take into account length of stay as well as bed day costs: treatment episode cost = (length of stay in days) x (bed day cost)
In 2009 Brown and Fahy studied 157 male patients discharged from medium secure units; their report, Medium Secure Units: pathways of care and time to discharge over a four-year period in south London, showed they found those on restriction orders had much longer admissions than those on other sections, and that discharge was often delayed due to a lack of supported mental health hostel placements in the community. Neither of these factors is within the immediate control of treating clinicians.
The median length of stay for study participants was 720 days or 2.7 years (interquartile range: 0.8-3.5 years), demonstrating the wide variability in length of stay and therefore also cost. The median treatment episode cost, based on £451 per day, would have been £325,000.
Due to patients’ diversity, the average length of stay for the entire population is less meaningful. A classification system of five forensic pathways allows various types of cases to be differentiated based on their presenting characteristics and projected care package needs. This system was developed at the Bracton Centre medium secure unit, part of Oxleas Foundation trust in London and in other medium secure units in the capital over the last two years and has been refined and used in several units nationwide. It is easy to use, acceptable to clinicians and quick (if the patient is well known to the clinician, it takes about a minute per patient).
Fromeside is an 80-bed medium secure unit in Bristol; 68 of its beds are for men. The 133 men discharged from Fromeside between 2005 and 2011 had a mean age of 35 years (range: 18-62 years) and most were on sections 37/41, 47/49 or 48/49, with around a quarter of patients on each section. Most patients had committed a violent index offence or a property offence (71 per cent and 20 per cent respectively), 71 per cent had a primary diagnosis of psychosis, 15 per cent a primary diagnosis of personality disorder and 6 per cent had a primary diagnosis of affective disorder. The remainder had other diagnoses including dementia and developmental disorders. Almost a quarter of the patients had multiple diagnoses.
Over half of patients were admitted from prison and 28 per cent were admitted from other medium secure services. On discharge, 32 per cent went into the community, 22 per cent were transferred to prison and 23 per cent were discharged to low secure services. The remainder were discharged to high or medium secure services and one patient died.
The occupied bed day cost for Fromeside is above the national average (£572 per day) and the median length of stay is 296 days (mean 461, range: four to 2,374) compared with a median of 720 days in the London study. The median treatment episode cost for Bristol, based on a bed day cost of £572 and median length of stay of 296 days, is £169,300 compared with the figure of £325,000 calculated from the London study.
Medium secure units operate in the political context of crime and public safety, so there are also societal costs to consider for unsuccessful unit outcomes such as reoffending and readmission. Readmission, it is worth noting, may be a positive and a negative outcome because, despite a failure in the community, such an intervention (recalling a patient) may prevent harm.
However, a reduction in bed day cost could increase treatment episode cost and result in a worse experience for patients. Just as a higher bed day cost is likely to be linked to higher staffing and better services, a lower bed day cost may reduce quality by cutting staff, leading perhaps to fewer therapy and escorted leave opportunities; these in turn will increase length of stay.
Current methods for measuring and comparing treatment costs are problematic for medium secure units; occupied bed day costs are misleading because they disregard length of stay. Patients are misrepresented within the existing 21 mental health clusters because individual clusters do not reflect the complexity of problems and treatments required. These patients have complex needs, are socially excluded and misunderstood so it is appropriate that they receive a range of treatments and timely discharge from the expensive units to which they are admitted. This supports the principle of patients spending the minimum amount of time in the least restrictive environment which is not only best clinical practice but also has the benefit of minimising costs.
We propose that patients in medium secure units are categorised according to the five forensic care pathways. These have already been used in practice in several medium secure units and are easy to understand, quick to use and reduce bureaucracy. This system is a parsimonious solution, providing a balance between simplicity and coverage (only 7 per cent of our sample could not be allocated to a forensic pathway by their consultant).
The five forensic pathways allow treatment episode cost to be calculated for particular groups of patients, which is more useful than calculating an average treatment episode cost for all patients in medium secure units. This approach also allows service providers and commissioners to gain a better understanding of the full economic costs.