Lean thinking

Getting to the essence of ‘lean’ is made challenging by the ambiguous language and definitions in the field. Hallam captures this well in observing that the term ‘lean’ has been used interchangeably to describe four different dimensions, specifically ‘the operating philosophy, the tools, the activities, and the state of the manufacturer’ (cited by Stone^{Reference Stone1}). In his review on the science of lean, Kyle Stone defines the lean thinking paradigm as the ‘operational philosophy’ of the organisation’ which ‘differentiates between waste and value’.^{Reference Stone1}

Muda (Japanese for waste) is defined by Womack & Jones as ‘any human activity which absorbs resources but creates no value’^{Reference Womack and Jones2} (p. 15). The ‘powerful antidote’ to tackling waste, they argue, is lean thinking. Five ‘lean principles’ are proposed. Data, evidence and metrics, and their application are not specifically addressed within the principles, but their use in realising the principles is clearly important. Definitions of the principles exist, but there are challenges in engaging with them and applying them to healthcare. The next section considers these challenges in relation to the key concept of each principle.

Challenges engaging with the lean principles

Lean in practice: a case example

Tees, Esk and Wear Valleys NHS Foundation Trust (TEWV) were supported by Virginia Mason in the USA in adopting lean.⁸ Virginia Mason's own lean processes were adapted from the Toyota Production System, with a strong focus on eliminating waste (muda) and the use of metrics to measure improvement.^{Reference Patterson9} TEWV have made a number of high-impact changes to the way their processes operate that have had benefits (brought value) to patients. One of these, the purposeful in-patient admission (PIPA) model, which was trialled on two adult wards, saw their original bed occupancy of 106% reduce by 22%, a 57% reduction in length of stay, a 72% reduction in reports of violence and aggression, and a 100% reduction in complaints (p. 62).^{Reference Crisp, Smith and Nicholson10}

Lean techniques they used to achieve these changes included:

• • a move from a weekly ward round to daily multidisciplinary team meetings – to remove ‘batching’ of decision-making (to improve flow of the value stream of clinical decisions and interventions)

• • visual control boards on the wards – to map the patient journey (flow)

• • removal of waste (muda) from the office and patient literature – ensuring that only necessary literature were on the ward

• • creating standard processes for each step of the patient journey and changing the layout of the ward environment (to improve flow).

The Health Foundation is now exploring how other organisations might benefit from lean and is conducting research into the effect a partnership with Virginia Mason would have on five other NHS organisations.¹¹

Lean thinking – evidence of impact

Although there are countless books and journal articles on the subject, Stone^{Reference Stone1} identifies three ‘voids’ that lean research is yet to adequately tackle:

1. (a) the relationship between planned organisational change and human resource development interventions

2. (b) the relationship between planned organisational change theories and lean theories

3. (c) the ‘human’ factor – by this he means how best to engage staff in lean.

In contrast, in The Toyota Way, Liker defines the tenth of the 14 management principles of Toyota as ‘develop exceptional people and teams who follow your company's philosophy’^{Reference Liker12} (p. 184) and explains that at the company the Toyota Production System was originally called the ‘Respect for Humanity System’^{Reference Liker12} (p. 186). Nevertheless, other than his own observations, and a critique of theories of behaviour, Liker offers no empirical evidence for human resources and the human factor.

The Model for Improvement

Don Berwick, the improvement champion who has led the Institute of Healthcare Improvement (IHI) in the USA and who was called on by the then Prime Minister David Cameron to help make the NHS safe for the future¹³, writes an impassioned introduction for The Improvement Guide, a key text that describes the Model for Improvement.^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} Notably though, the authors of the guide, who Berwick says he is still learning from, are largely statisticians by training. This is important, as the focus in lean is on waste and flow, whereas data, and the use of data, are at the heart of the Model for Improvement.

Berwick calls the model the ‘most useful single framework’ he has encountered^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} (p. xiii). Founded on the work of Deming, also a statistician, the ‘System of Profound Knowledge’ is fundamental to the Model of Improvement. The system asserts that, through developing appreciation for a system, understanding variation (data, evidence and metrics are clearly important here), building knowledge and getting to grips with the human side of change, one is better equipped to make improvements^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} (p. 76). Applying the Model for Improvement to an improvement challenge helps the leader organise an approach by guiding them through the following key questions.

1. (a) ‘What are we trying to accomplish?’ is about being specific and defining the problem. This makes it measurable and potentially subject to metrics. The question is also used as a call to arms, by clearly describing what is expected, and to create ownership, through encouraging others to contribute to creating this definition or ‘charter’^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} (p. 89). There are some darker sides to how this statement is used to motivate change, of the stick rather than carrot variety. For example, Langley and colleagues^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} suggest choosing goals that are clearly unachievable using current practices, being explicit that previous tools will be removed whether change happens or not, or simply stating that the current service will no longer be provided if change does not happen. Thus, although values of addressing the ‘human side’ of change are espoused, the authors are not shy of using traditional ‘machine metaphor’ classic top-down management theory^{Reference Morgan15} (p. 18) if they think it will help towards achieving the goal. This seems to be in contrast to the approach Berwick advocates in his work on patient safety for the NHS, where he says, ‘Fear is toxic to both safety and improvement’.¹³

2. (b) ‘How will we know that a change is an improvement?’ is sometimes abbreviated to ‘measures’. This is all about how to demonstrate the impact of change using data, evidence and metrics. Three different types of measures are encouraged: outcome measures that observe the outcome in question, process measures that monitor whether activity to achieve the outcomes is performed, and balancing measures that look at whether there are any unintended consequences of change^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} (p. 96). One could argue that the model encourages only superficial engagement with the ‘is the change an improvement?’ part of the question by focusing on measures. For example, in a healthcare system where increasing discharge is the aim, is there enough challenging of the assumption that discharge is the right thing?

3. (c) ‘What change can we make that will result in improvement?’ is often abbreviated to ‘changes’ and is about identifying initiatives that could bring about change^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} (p. 93). Methods for developing change are promoted, including ‘logical thinking about the current system, benchmarking or learning from others, using technology, creative thinking and using change concepts’^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} (p. 120). ‘Changes’ can be opportunities to put evidence into practice, and could be an application of evidence-based medicine, such as the implementation of National Institute for Health and Care Excellence (NICE) guidance.¹⁶ Dozens of change methods are proposed that can be used in the Model for Improvement and many of these, such as ‘use pull systems’, ‘eliminate things that are not used’ and ‘match the amount to the need’, have clear roots in lean thinking^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} (p. 358). Langley and colleagues acknowledge the overlap with other improvement approaches: ‘Several of the concepts are included in other approaches to improvement, such as Total Quality Management, Reliability, Safety, Six-Sigma, and Lean’^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} (p. 358).

With the three Model for Improvement questions answered, improvers are ready to make use of the ‘plan–do–study–act’ (PDSA) cycle. PDSA can be used to ‘turn ideas into action and action into learning’^{Reference Langley, Moen, Nolan, Nolan, Norman and Provost14} (p. 97). There are four distinct phases to PDSA:

1. (1) the intervention or test should be planned

2. (2) the plan should be executed and data recorded

3. (3) data are analysed

4. (4) reasonable action is taken on the findings (essentially action is based on evidence).

The cycles of PDSA can be used to ‘build knowledge’ both of the improvement challenge faced and potential solutions.¹⁷ PDSA cycles are recommended by NICE to bring about improvements through implementing NICE recommendations.¹⁶ A model akin to ‘plan–do–study–act’, called ‘plan–do–check–act’, has been used in lean in, for example, work on patient safety.^{Reference Black, Miller and Sensel18}

The Model for Improvement in practice: case examples

East London NHS Foundation Trust (ELFT) adopted the Model for Improvement, supported by the IHI. The trust's work to reduce violence on in-patient wards saw a 40% reduction in violence across six wards and reduced costs related to violence by £181 296 (data are for 2015–2016).¹⁹ The Care Quality Commission (CQC), the UK's healthcare regulator, has rated the organisation as ‘outstanding’ and commented:

The teams used the Model for Improvement questions to define and drive their work. They agreed what they wanted to accomplish (to reduce physical violence by 30%) and the measures to determine whether a change was an improvement (the main outcome measure used was ‘rate of incidents of physical violence per 1000 occupied bed-days’). In generating ideas for change, the team worked with staff and patients, and then used PDSA cycles with wards from across the trust to test the favoured change strategies. Staff came together at 6-weekly intervals to learn from each other and review data to consider whether improvement was happening.^{Reference Taylor-Watt, Cruickshank, Innes, Brome and Shah21}

Hertfordshire Partnership University NHS Foundation Trust has also employed the Model for Improvement, launching it in 2015. Subsequently, the 2016 national NHS staff survey reported that the percentage of staff in the trust who said they were able to contribute to improvements had increased to 76%, from 70% the previous year. The Picker Institute, which analyses the data for the NHS, reported this as a statistically ‘significant improvement’ and above the national average for mental healthcare.²² The trust was also rated ‘outstanding’ by the CQC in May 2019. It said:

Model for Improvement – evidence of impact

Evidence for PDSA in a UK healthcare context is varied. A systematic review of quality improvement methodologies carried out by NHS Scotland found that the evidence for PDSA in the NHS was ‘mixed’. It found that, in large projects taken forward by the NHS Modernisation Agency, work often did not proceed beyond ‘plan–do’. Other initiatives, such as work on the 4-hour emergency department (A&E) waiting target, could lead to problems elsewhere in the system (such as the patient waiting in a medical assessment unit instead). Success can be context dependent, with ‘striking differences’ between organisations using PDSA in quality improvement collaborations.^{Reference Powell, Rushmer and Davies24} Furthermore a multisite cluster-randomised study comparing the effectiveness of different change initiatives within healthcare services in the UK found no difference between standard dissemination of guidance versus standard dissemination plus PDSA in the implementation of the guidance.^{Reference Rycroft-Malone, Seers, Crichton, Chandler, Hawker and Allen25}