Physical intervention

Restraint data were extracted from the Trust incident reporting system and compiled into monthly reports for the Positive and Proactive Care Review Group. The baseline figure for total number of restraints for 2014 (calendar year) was 328.

As explained above, during the first quarter of 2015, CPFT’s incident reporting system was refined; thus, the next two  12-month cycles map on to the financial year (April–March).

From table 1 and figure 1, we see a broad decreasing trend in both the number of physical intervention and prone restraint incidents, particularly in comparison with the 2014 incident numbers. Tests of a polynomial cubic trend within Poisson regression models for the both number of physical intervention incidents (excluding patient A; see below) and prone restraints were statistically significant (χ²(3)=78.86, p<0.0001 and χ²(3)=49.13, p<0.0001, respectively).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Restraint incidents (with and without a patient who was restrained many times; see text), and number of incidents involving prone restraint by month. Smoothed trend lines are produced by Friedman’s ‘super smoother’.

Prone restraints, a primary focus particularly early on in PROMISE, dropped from 82 (2015–2016) to 32 (2016–2017), a reduction of 61%. Restraint type used was not consistently recorded, but we expect the vast majority of the 328 incidents to be prone as this was standard practise then. Thus, we expect there was an even bigger proportional reduction from 2014 to 2015–2016.

Restraint incidents decreased from 328 (2014), 241 (2015–2016) to 210 (2016–2017) successive reductions of 27% and 13%. However, during 2016–2017, we had an example of an individual patient significantly impacting figures: the spike in May 2016 (Figure 1) was driven by patient A (online supplementary table 4 and figure 1) struggling with a severe eating disorder and a precarious body mass index; every feeding attempt was crucial but, unfortunately, required restraint to be successful. Excluding this patient, the total for 2016–2017 would have reduced to 124 and would have represented a 48% reduction on the previous year (2015–2016).

The overall reduction in restraint incidents shows the reduction in prone restraints did not result in a shift to other forms of full physical intervention (eg, supine, kneeling and sitting). A similar reduction over time was noted in the number of individuals being restrained (online supplementary figure 2).

The most marked improvements were observed in the specialist wards (table 1). For example, the learning disability (intellectual disability) inpatient unit had a large reduction in restraints from 51 (2015–2016) to 9 (2016–2017), an 82% reduction, and prone restraints from 9 to 2. A similar reduction was seen in psychiatric intensive care unit: 43 (2015–2016) to 8 (2016–2017), an 81% reduction, and prone restraints from 13 to 2. The specialist personality disorder unit was prone restrain free for 23 of the 24 months, and full physical intervention numbers fell from 11 to 1.

By March 2016, 16 of the 20 wards had a 6-month stretch of zero prone restrains, which increased to 18 out of 20 by March 2017. All 20 wards had a 4-month stretch of zero prone restrains in each year. For full physical intervention, by March 2016, 17 of the 20 wards had a 3-month stretch of no full physical intervention, which increased to 20 by March 2017.

Patient experience

Figure 2 shows the monthly ‘Overall Performance’ (a summary across items) of patient experience from April 2014 to March 2017. An average score of 87% for the whole period was recorded across 4591 surveys (approximately 75% of inpatients). There is a small significant downward linear trend (linear trend test in a linear regression model: F_34,35=28.63, p<0.0001). A key driver of this trend is thought to be a change in time of surveying to capture the experience of acutely unwell patients. 

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Monthly ‘Overall Performance’ for patient experience ratings (range between 0 and 100). Overall performance is a summary measure based on responses to individual items. Smoothed trend line produced by Friedman’s ‘super smoother’.

The top three items are attitudinal (staff polite and friendly: 98%; admission welcoming: 97%; respect and dignity maintained: 96%) and are followed in rank order by action oriented measures (medication purpose explained: 94%; weekday activities supported: 93%; have a care plan: 93%). The individual patient experience items are plotted by month in online supplementary figure 3 and reported by year in online supplementary table 2. To combine attitudes and actions, in August 2016, a new question was introduced to elicit whether patients felt involved in their care/treatment discussion. A 98% rating across 964 surveys on this item shows the effort staff made to involve patients in their care irrespective of their voluntary or involuntary status (online supplementary table 5).

Relationships between restraint numbers and patient experience

Patient experience data are completely anonymous and cannot be linked to individuals restrained. The correlation between the detrended (series values minus Friedman’s ‘super smoother’ fit, for April 2015–March 2017) overall patient experience and restraint incidents (excluding patient A) is 0.14, a small and non-significant correlation (p=0.5235). It is necessary to detrend the series to reduce the risk of spurious correlations caused by trend.

Lessons and limitations

At the heart of PROMISE is an alternative approach to governance. Initially, as an organisation, we were focused on targets and trajectories. These are important tools for measuring progress. However, engagement exercises highlighted that behaviours and practices that we were endeavouring to change through KPIs were only the visible tip of an iceberg. One could hit the target but completely miss the point. What kept restrictive practice embedded was the existing culture, the invisible bulk of the iceberg. We made a decision early on to grapple with culture and address it head on in order to avoid a situation where as soon as one target or behaviour is addressed another would emerge. This is tiring for leadership and demotivating for front-line staff. So cultural change24 was needed at two levels: first, front-line culture needed changing to reduce the use of restraint, and second, a wider cultural shift from focusing from ‘target chasing’ to a more holistic improvement mindset. We expected an improvement in the second would contribute to increased gains in the first.

The PROGRESS framework aimed to manage this transition of a target-driven governance mindset of ‘have to’ to an outcome based ‘want to’ approach through adopting temporal sequencing of the five actions. The former results in staff who are overwhelmed with delivering KPIs and are feeling defensive if they are failing. The latter empowers and enthuses staff to find novel solutions to challenges and draw on the existing expertise of patients and carers. The five actions of PROGRESS (Report, Reflect, Review, Rethink and Refresh) grapple with culture and create a mindful learning environment and ensure that the story of better patient care remains at the forefront. This is particularly important when addressing restrictive practice. Physical intervention does not sit well with core staff values of caring and was considered a necessary evil.2 Generally, staff are acting out of welfare or safety concerns and are at the receiving end of overwhelming distress in a patient. It is a highly emotive subject, and a governance framework that makes staff feel as if they are failing or being criticised will only make them defensive.25 Antecedents to restraint vary considerably, and defensive staff do not go looking for creative individualised solutions. On the contrary, energies get diverted to rationalising what is happening. Thus, a fine balance needs to be struck to ensure that the issue is not diluted, but simultaneously staff engage and own the improvement agenda.2 18 In PROGRESS, this is achieved through a natural counterbalance of summative data in the report and review actions and the formative non-judgemental curiosity of reflect and rethink steps. In fact, there is considerable overlap in the attitudinal aspects of all five steps.

The adult treatment wards provide an interesting case study. There were a large number of initiatives that the wards embraced that resulted in very high patient experience scores and a 33% decrease in 2014 restrain numbers, but there was limited decline in the next year. A careful analysis showed that most of the patients being restrained were those suffering from severe mania and needed restrain for their own safety (ie, to prevent absconsions or to protect their dignity). An atmosphere of more light (non-judgemental genuine curiosity) and less heat (to stop staff feeling defensive and demotivated) has been maintained throughout, so that the wards did not engage in target chasing (restraint numbers). Two key pieces of work are being undertaken. First, a study has been launched to assess whether advanced psychological formulation of predisposing, precipitating, perpetuating and protective factors can break the cycle of restraint in this group of very unwell hard to engage patients. Second, further qualitative analysis of electronic patient records of patients with repeat restraints to establish missed opportunities for early identification and intervention relating to escalating distress.

PROMISE proposes a complex multifaceted intervention, where culture is addressed as a means to change the nature of day-to-day interactions. Thus, it is difficult to attribute causality to any specific aspect of PROMISE. However, PROGRESS, with an alternative mindset to governance, provides a framework in which staff and patients can feel empowered to coproduce a new care culture  that moves practice away from coercion. It creates a therapeutic milieu in which practitioners can provide a caring response to distress, irrespective of its source or height.

The framework breaks the transformation journey into bite-sized chunks and makes it generalisable (online supplementary table 3). Different organisations will be at different stages in their journey of setting up systems and processes around quality assurance. Therefore, they may vary in the sophistication of their incident reporting systems, or culture of reflective practice, or ward to board accountability, or systems to support learning. Before adopting and/or adapting PROGRESS, we recommend a baseline exercise. Some quantitative data are needed to establish the scale of the problem relating to restrictive practice (eg, numbers of seclusion, restraints and rapid tranquilisation). If such data do not exist due to the absence of an incident reporting system, or it is not readily accessible, this would need to be prioritised.

Establishing a baseline for patient experience is equally important. Embedding systems and processes for routine reporting and reflection on patient experience and physical intervention data provided creative avenues to address restrictive practice. The monthly reviews were not just summative but also resulted in formative improvements to the survey (online supplementary table 5) and patient experience initiatives. The quarterly Rethink meetings with the wider leadership were focused on collating and sharing successes and challenges from these initiatives.

A proposal to raise the bar and capture the experience of patients when acutely unwell by changing the time of survey administration was finalised in the October 2015 annual Refresh (PROMISE Charter Event). In 2014–2015 surveys were conducted at discharge; subsequently, patients were surveyed as soon as clinically appropriate. This change is thought to be a key driver for the small downward trend. Additionally, following the national trend, CPFT has reduced its number of adult acute beds. Overall, this means that the average admitted patient will be more distressed and have higher severity of illness, resulting in lower patient experience scores. This effect is thought to be particularly acute in CPFT, with its reduction in beds of 44% compared with the national average reduction of 17% (Kay Ray et al, 2018).26 Within the context of these factors, restricting a change in patient experience to only a small decline is a substantial achievement. These issues highlight how patient experience is impacted by a great many internal and external factors: it is perhaps not surprising to find only a small relationship between restraint numbers and patient experience.

PROGRESS was supported by two active advisory groups (patients and staff) who provided in-depth insight into custom and practice. Coproduction was at the heart of this project, and the early involvement of patients and front-line staff played a significant part in the way governance structures iteratively evolved. As evidenced in  online supplementary table 6, PROMISE had a strong research strand from its inception to gather baseline cultural information.2 17 18 Research helped get the story out to the front line and created a stance of curiosity with a genuine desire to know rather than provide judgement. Contextual organisational research can supplement the foundations of a governance framework.