Embedding Recovery to Transform Inpatient Mental Health Care: The 333 Model

In the past decade, care for acute mental illness has been radically redesigned in a public-sector organization in the East of England. Catchment wards in Cambridgeshire and Peterborough National Health Service (NHS) Foundation Trust (CPFT) were replaced by a new model of inpatient care consisting of three time-limited inpatient pathways for assessment and specific interventions. Components of the model include 3-day assessment units (AUs), 3-week treatment units (TUs), and 3-month recovery units (RUs). The crisis resolution and home treatment teams (CRHTTs) assess patients prior to admission to the above units, provide home treatment as an alternative to inpatient stay, and facilitate early discharge. Rollout was conducted in phases, first in Peterborough (October 2011) and later in Cambridge (May 2013).

The model, known as 333, was designed as a systems solution to the universal problem of bed shortages. In the United Kingdom, the last two decades of the 20th century witnessed the closure of large mental health institutions and a move toward community care (1). In 1999 the National Service Framework was published (2). This was followed in 2000 (3) by a NHS plan that made CRHTTs mandatory for all catchment areas in England. There is now considerable evidence to suggest that CRHTTs are an effective alternative to hospital stay, allowing for bed reductions (4–6). In CPFT, implementation of CRHTTs allowed for closure of an inpatient ward in 2005 (see online supplement for a CRHTT patient flow diagram).

Despite the implementation of CRHTTs, inpatient stays for psychiatric care in the United Kingdom in 2016 lasted an average of 35.9 days, according to national benchmarking data (7). By comparison, inpatient stays for mental health care in the United States are considerably shorter, averaging 10.0±3.0 days for people with serious mental illness in 2006 (8), down from 25.6 days in the 1990s (9). These results reveal potential bed day savings in the United Kingdom, especially because the United States does not have an alternative to hospital stay comparable to CRHTTs. Many link the shortening of inpatient stays for psychiatric care in the United States to the economic considerations of insurance providers (10). However, in the United Kingdom, demand for beds has brought about a similar focus on patient flow, and the NHS has tried establishing 7-day assessment/short-stay units.

More recently the NHS has been under significant financial pressure, and organizations have needed to deliver on a steep cost-improvement program (11, 12). Simultaneously, heavily regulated stretch metrics have been enforced by both the quality (13) and the financial regulator (14). Hence, CPFT’s challenge was twofold, to push quality up and costs down. Improving quality of care and access to beds at a time when demand for beds far exceeded supply was not going to be easy. The bed crisis was driven by a decreasing number of beds (15), an increasing population from immigration into the east of England region (16), an increasing number of people in contact with mental health services (15, 17), scarcity of state-funded social housing (18, 19), and the impact of the Great Recession (20). Locally, timely beds were difficult to find, and 75 patients from the geographical catchment area of CPFT were being treated in long-term, private-sector out-of-area placements (OAPs) in October 2011. Inpatient treatment delays resulted in deterioration in the community. When admitted, patients were so unwell that they would need longer hospital stays and have poor outcomes, perpetuating the bed shortage cycle (21, 22). If they received an OAP, being away from their families resulted in poor patient experience (23, 24). OAPs diverted finances away from community services, which were delivering most of the components of the cost-improvement program (22).

However, reducing community services could be a false economy. In 2016, the cost per year of an adult acute bed stood at 131,267 British pounds ($161,879 U.S.), 6% higher than in 2015 (7). For an individual patient, a year of generic community mental health support cost 2,880 pounds ($3,551 U.S.), 3% higher than in 2015 (7). The cost of inpatient care increased more rapidly than did outpatient care, and annually the expenditure of one adult mental health bed can support delivery of care for 46 community patients. Decreasing community provision delays treatment, resulting in more unwell patients needing beds, deepening the demand-supply mismatch (22). As a result, the cost-improvement program was redirected from community to inpatient services.

The 333 model was designed as a systems solution to inpatient care delivery. The goal was to decrease OAPs while addressing a reduction in the stock of generic adult mental health beds to meet the necessary reductions in cost base. It was also hoped that the new model would allow for the release of funds from inpatient care, permitting them to be reallocated to make community services robust and responsive.

The 333 model sought to bring about a significant improvement in quality of care, with a specific focus and specific functions for each unit (Box 1). By providing clarity about the added value of every inpatient hour, the model’s recovery-focused approach was expected to release bed days, make inpatient care timely, result in rapid recovery with better outcomes, and shorten lengths of stay.

This article describes the performance of 333 on various process measures compared with national benchmarking and targets set at the model’s inception. To our knowledge, this is the first publication that attempts to provide evidence for the effectiveness of a systems solution to the universal problem of bed shortage. A literature review did not reveal any articles that report on the effectiveness of such approaches. This gap in the evidence base urgently needs addressing.

Methods

The complement of general adult beds in CPFT before and after implementation of 333 was mapped out. Raw data for 2 fiscal years (April 2015 to March 2017) were extracted from the trust’s data warehouse. These data were compared against 2016 NHS benchmarking (7) and targets set at the inception of 333. Data were categorized as admissions and patient flow, length of stay and readmission rates, and monthly occupancy levels. Under admissions and patient flow, targets included 1,200 care episodes annually, each equal to a patient journey across 333; a direct ratio of admissions to AU, TU, and RU of 75:25:0; discharge of at least 70% of patients from AU directly to the community; transfer of a maximum of 30% of patients from AU to TU, a maximum of 20% from TU to RU, and a maximum of 5% from AU to TU and RU; and a maximum backflow of patients from TU or RU of 5%. Targets for length of stay and readmission rates included a median length of stay of 3 or fewer days for AU, 3 or fewer weeks for TU, and 3 or fewer months for RU; a 30-day readmission rate of 15% or less; and a 7-day readmission rate of 5% or less. For monthly occupancy levels, targets included bed occupancy rate (70% or less for AU, 95% or less for TU, 100% or less for RU, and 85% or less overall, with a bed considered vacant if the patient had returned home for a visit) and nights per year without available beds (zero across the CPFT, ≤5 across AUs, ≤10 across TUs, ≤5 in Cambridge, and ≤5 in Peterborough , with a bed considered vacant if the patient had returned home for a visit).

Results

Bed Complement and OAPS Before and After 333 Implementation

Table 1 provides information about the changing complement of adult mental health beds from 2011 to 2016 within CPFT, which serves an adult population of 550,000 (16). Given that smaller wards are at the heart of healing environments (25), large catchment wards were functionalized and rebranded into smaller 333 units. These are supported by regional specialist units. Table 1 shows that after 333, there was a 44% reduction in beds in CPFT compared with a reduction of 17% nationally (7). Over 70 OAPs were repatriated, and new OAPs became extremely rare. When repatriation of OAPs is included, the percentage of bed reductions rose to 62%. As of December 2016, there were 16.9 adult acute beds in the CPFT per 100,000 adult population compared with a median of 19.9 nationally (7).

Admissions and Patient Flow

Table 2 shows that over the two years, 2,679 patients received inpatient care in the 333 system, exceeding the annual target of 1,200 by 11.7%. The target was set through a bed mapping exercise, assuming bed occupancy of 85%. The target ratio of admissions between AUs and TUs (75:25) was designed to permit the retention of generalist assessment skills on TUs, while allowing both units to specialize in their stated function. As detailed in the discussion, it was also a systems approach to decreasing restrictive care, one of the goals stipulated under the Mental Health Act. The target ratio was delivered successfully—76% of patients were admitted to an AU, 23% were admitted to a TU, and direct admissions to the RU were extremely rare (see the patient flow diagram in the online supplement).

Because median length of stay is a primary target, there was concern that it could become a singular focus, leading to repeated transfers and fragmentation of care. To mitigate against unnecessary transfers when discharge was impending, the percentage of discharges into the community from those admitted to AUs was set at 70%. Transfer to the next step was dictated by patient need and not by time elapsed since admission. Time scales of 3 days, 3 weeks, and 3 months were indicative and not absolute. Table 2 shows that 74% of patients were discharged back into the community from an AU and only 25% were transferred to a TU for further treatment. In a similar way, only 16% of patients on TUs were transferred to RUs. Only 1 in 23 patients was exposed to all three steps of the pathway, and only 1 in 26 patients (settled mover) needed to be transferred to an alternative bed to make room for another, more appropriate patient. All balancing targets set around patient transfers were delivered.

Lengths of Stay and Readmission Rates

Table 3 provides details on length of stay for each step of 333. The median targets of 3 days, 3 weeks, and 3 months were chosen on the basis of a bed mapping exercise that aspired to eliminate OAPs through bed availability. The AU target was missed narrowly, but the TU and RU targets were met quite easily. Across 333 the median length of stay was 5.83 (2015–2016) and 6.13 (2016–2017) days compared with a national benchmark of 36.1 days in 2016 (including days spent on leave from the unit) (7). The average stay in 333 units was 15.32 days in 2015–2016 and 16.02 days in 2016–2017 compared with 35.9 nationally (including days spent on leave from the unit) (7).

Seven- and 30-day readmission rates to any adult ward were 7% and 17%, respectively, in 2015–2016 and 5% and 13%, respectively, in 2016–2017, compared with target rates of 5% or less and 15% or less, respectively. Readmission is the anticipated consequence of short lengths of stay. However, continuous bed availability allowed for early readmission at first signs of deterioration, empowering patients and clinicians to embrace positive risks. On April 1, 2015, the AUs launched Open Door, a proactive initiative to manage care for patients with a severe personality disorder (26). The 333’s readmission rates for 30 days—13% to 17%—appear to be considerably higher than the national mean of 8.4%. However, removing Open Door readmissions for 10 patients with 10 readmissions each in 2016–2017 would cut the 30-day readmission rate for that year in half. This would bring the 2016–2017 30-day readmission rate at CPFT below the national average (7).

Bed Occupancy

Both AUs had a pivotal role in ensuring bed availability, and they consistently delivered on a stringent 70% target (Table 4). National median bed occupancy (excluding temporarily unoccupied beds) for 2016 was 94.2% (7) compared with 81% for CPFT, well within the overall occupancy target of 85%, a target in keeping with Royal College guidance (25, 27). National median bed occupancy (including temporarily unoccupied beds) was 102% in 2016 (7) compared with 88% in the CPFT.

Occupancy figures average out changes in pressure for beds and conceal peaks in bed demand. Given the aspiration to eliminate OAPs, an annual target of zero was set for number of nights without 333 bed availability (with beds of patients on overnight leave considered vacant). When necessary, beds of patients on overnight leave were used for new admissions. Within 333, patients were discharged daily and under rare circumstances when a bed of a patient on leave had to be used overnight, the ward team could find a bed the next day. This strategy helped reduce unnecessary OAPs. All site and function specific targets were met (Table 4). There was not a single night over 2 years in which a 333 bed was not available in CPFT. There were two nights when beds were not available at both sites. The only bed available on March 20, 2016, was in Peterborough, and on November 27, 2016, the only bed available was in Cambridge.

Discussion

The 333 model provides a solution to the national mental health bed crisis by offering time-limited pathways for assessment, treatment, and recovery and ensuring access to inpatient treatment should it become necessary. The Royal College of Psychiatrists recommends limiting occupancy to 85% in order to ensure quality and safety (25, 28). Rates that are 10% above recommended levels are associated with violent incidents on wards (29). Inspections have shown that seclusion rooms are sometimes used to accommodate patients who are admitted when a bed is not available (30, 31). OAPs are a natural outcome of this bed crisis. A freedom of information request found a total of 4,447 OAPs among 37 NHS mental health providers in 2014–2015, up by 23.1% from the previous year (32), and 88% of the OAPs were due to local bed unavailability. OAPs not only affect patient experience negatively but also have been associated with increases in patient suicides (20). The 2016 national benchmarking figures for occupancy demonstrate that overuse of OAPs has reached crisis levels (7). The 333 model was designed to counter this situation by delivering zero nights without beds annually. In addition, it supported a 44% bed reduction and the virtual elimination of long-term OAPs.

A Change in Mind-Set

The 333 model represents a radical shift in care delivery. In the lead-up to the implementation of the model, CPFT’s adult mental health leadership group identified mind-sets that limit innovation (33) and precepts for guiding actions to change them. For example, to open up silos, the precept was “Silos to collaboration: we are in this together.” Other aspirations included staff putting equal value on professional expertise and lived experience of mental illness (“Top to tap: I am here to help you help yourself”) and putting the person before targets (“Targets to outcomes: doing right by the person”). The hope was that this would create a care environment that was collaborative, recovery oriented, and person centered.

To realize this future, prior to implementation, five engagement events were held with over 300 stakeholders. A four-step influence model was used to guide the change journey: script a compelling narrative, set up reinforcement mechanisms, help staff acquire skills for change, and encourage leaders within the service to serve as role models for the new behaviors and mind-sets (33). At the engagement events, attendees defined “where we are” and shaped “where we want to get to and how.” They participated enthusiastically in open and honest discussion about what’s in it “for me, my patients, my team, my organization, and society” (34). Site visits to various organizations and detailed process/bed mapping were carried out. During target setting, as discussed earlier, balancing measures were incorporated to reinforce the patient’s story and disincentivize chasing targets, particularly length of stay. Because collaboration was the focus, system performance took priority over unit-level targets.

An open redeployment process was conducted to ensure that the skills and interests of key individuals—such as consultant psychiatrists, modern matrons (senior nurses), and ward managers—were aligned with the pathways to which they were assigned. This was to ensure that they could act as role models for the rest of the staff. In a tiered fashion, this process of realignment was replicated throughout the 333 workforce. There was a degree of short- term uncertainty but once completed, it ensured that the workforce was fit for purpose and staff felt passionately about their new roles. To ensure there was adequate time to train staff and promote new attitudes and aptitudes, one ward was temporarily closed. A majority of the staff group on this ward was going to take up roles in the new AU and they received intensive training in assessment skills. The same staff group was then used to backfill staff on the other wards so they too could be released to train in the skills that would be needed on a TU and RU. This approach minimized staffing disruption to ongoing care provision and simultaneously ensured that training was not overlooked. Some training was delivered by patients, a crucial step in shifting the clinician’s role from being “on top,” the paternalistic stance typical of mental health treatment, to being “on tap,” or coaching (35). Valuing the patient as an expert in the management of his or her condition was a major paradigm shift.

Embedding Recovery

The recovery principles of finding and maintaining hope, reestablishing a positive identity, taking control and responsibility, and building a life meaningful to oneself are central to the 333 model (35). For example, in the 3-day AU, staff consider themselves hope vendors. Without question, 3 days is inadequate to treat a person who is severely unwell. However, 3 days is ample time to initiate the journey of rediscovering hope and establishing a deeply empathic relationship. Within this scaffolding of relational safety, patients rediscover that all is not lost. This is the foundation of the safety planning balance sheet, which involves a process of shared risk response and mitigation. Even though patients continue to need treatment, the median length of stay in the AU is 3.75 days and most patients (74%) are discharged to home, where treatment can be safely managed with recovery-oriented interventions. There were fewer suicides among patients in CPFT (median=5.5, according to CPFT’s National Confidential Inquiry safety scorecard) than among persons reported by the National Confidential Inquiry to be under mental health care in 2013–2015 (7.1 per 10,000) (36).

Added value from a stay on any ward is scrutinized with regular exploration of alternatives. Working in partnership, staff and patients establish realistic yet challenging recovery milestones toward which they work together and hold themselves to account. Each pathway has well-defined roles. For example, in addition to the core function of “assuring safety through hope,” AU staff have five specific assessment goals. These include a formulation of the patient’s needs and strengths, a provisional diagnosis, what treatment is indicated, where is it best delivered, and for how long. Every hour of every stay is accounted for on the AU—with clear milestones for every hour, beginning with hour 1. The assessment process is continuous and is integrated into the daily routine. A patient’s attempt at a jigsaw puzzle creates concentration-related documentation. If peers are interacting, staff observe for eye contact, rapport, and reactivity of mood. This continuous process feeds into a holistic formulation that guides the opinions relating to treatment and safety. Like the AU, the TU has day-by-day milestones, whereas milestones in the RU are measured week by week.

Common sense as common practice was adopted by staff as a principle to navigate the interfaces in the patient journey. This was central in putting the patient first. Patients do not fit into pathways; rather, the pathways fit around the patient. In 333, clinical judgment and common sense overrule archetypal policies and protocols. Dialogue is the preferred way to resolve any pathway bottleneck. To provide seamless care, key individuals, such as the community care coordinator, remain involved all through the inpatient stay. The same consultant psychiatrist would follow the patient across TU and RU, providing consistency and continuity of care. Similarly, the CRHTT and AU work very closely and are collocated, sharing staff and expertise. This provides a smooth transition from CRHTT to AU and back to CRHTT on discharge, eradicating wasteful double assessments.

New Initiatives

The new mind-sets have created a can-do attitude and a milieu in which innovation centered in “doing right by the patient” flourishes. For patients admitted through the Open Door initiative, for example, the AU serves as a safe, short-term haven in which to allow distress to settle (26). In the past, to manage risk, a patient with a personality disorder would sometimes be transferred to a TU, where long stays would achieve little. Research shows that a lot of resources are unhelpfully spent in avoiding hospitalization of patients with a personality disorder following presentation to an emergency room (37).

To avoid this outcome, staff proposed the Open Door program, a radically opposite approach that involves taking down barriers to admission and handing control over to the patients. As long they have not engaged in self-harm, patients in the Open Door program can request a 2-day admission without having to give any justification. Following discharge, patients are expected to remain at home for 24 hours before making a further request for admission. The therapeutic benefit of the approach lies in the relational safety constructed between the staff team and the patient through a sequence of assured Open Door admissions that validate the intense psychological pain when in suicidal distress. The safety net creates a sense of containment and admission requests space out fairly quickly, as most experience long periods in which they are free of self-harm. Individuals treat the privileged access with extreme responsibility, and admission requests spaced out fairly quickly. Most experience long periods in which they are free of self-harm, and Open Door privileges are withdrawn extremely rarely. An initiative like this could be launched only in a system in which beds are reliably available. Although Open Door doubled CPFT’s readmission rates, it changed the care-seeking trajectory of these patients. They were proactively seeking help and getting the necessary support rather than presenting in a crisis and then having to fight against the system to justify that their distress needed containment in an inpatient setting. The leadership took the view that there is little to be gained by hitting a target but completely missing the point of putting patients first.

The 333 system design reinforces recovery principles. For example, patients admitted on an involuntary basis bypass AUs and are admitted directly to the TUs. Recovery-oriented care is incentivized by setting the target ratio for direct admissions to AU and TU at 75:25. In 2011 the ratio of voluntary to involuntary admissions in adult mental health in CPFT stood at 65:35. Thus the ratio of 75:25 reflects an organizational aspiration to provide the least restrictive care by increasing the proportion of patients admitted to the AU, which would have the effect of increasing the number of patients admitted voluntarily. Invocation of the Mental Health Act declined because continuous bed availability meant that patients received timely inpatient care before they became extremely unwell in the community. Also, for many reluctant patients or those worried about the stigma of a hospital stay, a 3-day voluntary admission was a compromise they were willing to make. National benchmarking data from 2016 showed that the mean proportion of patients admitted involuntarily was higher on a national basis (35.1%) than in the CPFT (19.8%) (7).

Over the same period, a parallel program to decrease coercion in care (PROMISE) was successfully implemented, and physical interventions by 36% and prone restrains by 58% (38). Inpatient experience surveys (N=4,591) administered between April 2014 and March 2017 provide further evidence of high-quality care: 87% rated the care received as good, very good, or excellent, with 98% feeling involved in their care or treatment. Staff attitudes were highly commended (staff polite and friendly, 98%; admission welcoming, 97%; respect and dignity maintained, 96%), as were action-oriented measures (medication purpose explained, 94%; weekday activities supported, 93%; have a care plan, 93%). High scores were registered in spite of the involuntary status of approximately 20% of patients (38).

Limitations

The 333 model provides a viable solution for managing the resources available for inpatient mental health care; however, it does not meet the needs of patients who require longer-term care and must have support from specialist wards. The system is interdependent, and any issue in one part affects the whole. Without daily oversight, patient flow tends to break down across interfaces. Positive risk management is crucial to its success, as is an engaged, recovery-oriented workforce that believes in the model. Thus recruitment, training, and retention of skilled and experienced staff are essential.

This article provides a range of process measures and makes reference to patient safety and experience. Claims of high-quality care are based on enhanced access, shorter lengths of stay, more treatment in the least restrictive environment, and decrease in OAPs but not on clinical outcomes. The primary outcome measure used in CPFT for a finished pathway episode is the Health of the Nation Outcome Scales (HoNOS) score (39). HONOS is administered at initial entry into CPFT and repeated every 6 months and at discharge from secondary care mental health services into primary care. HONOS clusters are linked to the “payment by results” initiative in the United Kingdom, and a drop in HONOS scores provides valuable information for the overall patient journey through NHS trusts. However, HoNOS is scored on the basis of symptoms in the previous 2 weeks, making it unsuitable for assessing outcomes for most patients in the 333 system, where median lengths of stay (AU=3.71 days and TU=15.12 days) are too short to capture changes between admission and discharge. To address this issue, attempts were made in late 2017 to incorporate the Clinical Global Impressions Scale into standard practice for 333 (40). However, reporting on that is outside the scope of this article.

Conclusions

With its recovery-focused approach, 333 has reduced lengths of stay and ensured that a stay on any ward is meaningful and adds value. The bed capacity that has been created has made the acute care service more responsive, resulting in earlier intervention, decreased suffering, and improved outcomes. Over 70 patients in OAPs have returned home for treatment, and bed stock has been reduced by 44%, a reduction that has released funds for the cost improvement program and reinvestment in the community. The new mind-sets promote collaboration through common sense and add value to individual recovery journeys. On most metrics, 333 has outperformed national figures and has now developed a track record of delivering high-quality, seamless, cost-effective, safe, and innovative care. As this article demonstrates, reductions in beds and OAPs can be achieved simultaneously with the delivery of safe, high-quality care.