Patient safety near misses – Still missing opportunities to learn

Objectives A patient safety near miss is a safety event that had the potential to cause harm, but did not reach the patient. For over 20 years healthcare has been exhorted to learn from patient safety near misses to support improvements in patient safety. The belief is that, by addressing the factors that contribute to patient safety near misses, harmful incidents will be avoided. However, there seems to have been little progress made to learn from patient safety near misses. This study aimed to explore why there has been limited progress, and how best patient safety near misses may be learned from. Methods A qualitative case study was undertaken to explore the learning from patient safety near misses in different National Health Service contexts. Semi-structured interviews were conducted with patient safety leads in secondary care, primary care, and regional/national bodies. Interviews were recorded, transcribed, and thematically analysed. Results Seventeen interviews were undertaken across the National Health Service, with further data collected from policy, guidance, field notes, and research memos. Thematic analysis identified the following: variations in safety event schema; limited processes for patient safety near misses; unsupportive reporting contexts; and assumed, but non-evidenced improvements in patient safety. Participants also shared their thoughts on how learning from patient safety near misses could be improved. Conclusion A lack of progress has been made to learn from patient safety near misses in the National Health Service. This is contributed to by a lack of agreement about what is and how best to learn from a patient safety near miss. The learning value of patient safety near misses lies in the focus they place on controls to hazards, but they should not be learned from in isolation.


Introduction
2][3][4] In England, the [former] Department of Health (2002) described that 'No account is taken of near misses'. 2 Now, over 20 years later, the literature suggests that healthcare is still missing out on opportunities to learn from those near misses that occur during patient care 5 ; referred to as patient safety near misses (PSNMs) in this study.
The World Health Organization (WHO) describes a PSNM as an '…incident which did not reach the patient'. 6PSNMs are valuable because they are common, provide learning to support safety improvements, and are of low emotional impact. 4,7,8The theory is that, because contributory factors to PSNMs are the same as to incidents, 9 they are 'free lessons' that will help prevent incidents. 4ther industries, such as aviation, have 'evolved incident reporting systems that focus on near misses' and have found value in learning from them. 10 In those industries learning contributes to system resilience. 11hile the benefits of learning from PSNMs are stated, healthcare may not be effectively exploiting that learning. 5o support learning from PSNMs the authors set out to study their management in the English National Health Service (NHS).The questions asked were (1) Why has there been limited progress to date in learning from PSNMs, and (2) how best may they be learned from?

Methods
A Qualitative Case Study (QCS) was used and followed a published approach. 12QCS supports the exploration of 'how and why' questions around a specific phenomenon (PSNMs) in context (healthcare sectors). 13Before commencing the QCS the following propositions were stated: there is a standard definition of a PSNM in the NHS; PSNMs are included in organisational governance processes; and learning from PSNMs improves patient safety.

Cases and sampling
The sectors chosen as representative of the NHS were secondary care (acute hospital, mental health, and ambulance), primary care (general practice), and regional/national bodies (regulatory and policy).The selection of secondary care organisations was guided by Care Quality Commission rankings for safety to seek high-performing organisations.For primary care, commissioning groups familiar with local performance were consulted.Patient safety leads (participants) in organisations were approached.This purposive approach created a sample that could be assumed to be knowledgeable about PSNMs in their organisations. 14

Data collection
Participants took part in semi-structured interviews.One researcher undertook all interviews for consistency.Interview questions (Supplemental Material) were refined with nine safety professionals before piloting.Further data included field notes, research memos, and guidance/ policies.

Data processing and analysis
Interviews were audio recorded and transcribed.Analysis was undertaken using qualitative software and commenced following the first interview, continuing alongside data collection.The analysis followed guidance and a logic model was used to support the analysis. 15,16Logic models link underlying theoretical assumptions with processes and outcomes. 17As no pre-existing logic model for learning from PSNMs was found, a draft model was developed (Supplemental Material).
One researcher undertook initial coding of the data to identify preliminary themes against the logic model.Data and preliminary themes were reviewed amongst the researchers to refine the themes.All researchers had experience in thematic analysis.

Enhancing trustworthiness
Research and ethics approvals were granted prior to commencing.Relevant standards were followed to support rigour. 18,19Trustworthiness was created through the protocol, stating of propositions, the logic model, and triangulation of data and perspectives.Draft themes/findings were shared with participants for verification, and the researchers engaged in reflexivity throughout.

Results
Data collection took place between 2019 and 2021.Sampling continued until themes were identified across the healthcare sectors with agreement amongst the researchers (theme saturation).Seventeen interviews (Supplemental Material) were undertaken across acute hospitals (AHs, n = 3), mental health organisations (MH, n = 2), ambulance services (ASs, n = 2), general practices (GPs, n = 6), and national/regional bodies (NBs, n = 4).Twenty-eight documents were included.Final codes and a revised logic model (Supplemental Material) were used to develop the thematic map (Figure 1).Example quotes in support of the themes are provided in Table 1.

Variations in safety event schema
While the concept of a PSNM was recognised by all participants, the majority were not aware of a standard definition.Participants stated their own definitions which showed variation in what was interpreted as a PSNM.The various definitions included events that were interrupted before reaching a patient, events that reached a patient but caused no harm, and events that reached a patient and caused harm (albeit less).Multiple participants thought an 'interruption' was needed for a situation to be a PSNM as opposed to a 'no-harm incident'.
Multiple participants, including a national body, conflated different events under the title of PSNM.These included events that had been interrupted before reaching a patient, and events that had reached a patient.Their rationale was that these all provided the same learning.However, other participants questioned whether interruptions in events prior to reaching a patient provided greater insight into the presence and effectiveness of actions to prevent harm.

No specific processes for near misses
No specific processes for learning from PSNMs were found across the organisations.This meant, much like was described for other no-harm events, there was limited or no investigation, or learning.Participants said this loss of learning was the 'norm'.A small number described sporadic investigations for PSNMs where there was a risk of external scrutiny.Investigations focused on contributory factors and did not explore interruptions.
A review of local safety policies found reference to PSNMs in secondary care organisations.Policies often stated the need to report 'all incidents and near misses'.However, participants described the PSNM inclusion as 'tokenistic', with no reference made as to why or how they should be learned from.

Unsupportive reporting contexts
All participants described underreporting of PSNMs in their organisations and did not know how many occurred.Underreporting was felt to be due to limited encouragement, accessibility issues with incident reporting systems (IRSs), non-supportive safety cultures, and competing demands.No local or national direction supporting the reporting of PSNMs was found, and PSNMs were rarely discussed during induction or mandatory training.It was also stated that national bodies provided 'mixed-messages', and one participant recalled being told 'not to bother [with PSNMs]' by a national body.
Safety cultures were felt to influence the reporting of PSNMs.Secondary care participants were not confident that their cultures supported staff to report, and described blame originating from leaders, colleagues, and the media.In primary care, while blame was still described, participants felt supportive cultures were easier to develop due to having small and often long-serving teams.
All participants believed in the need for healthcare to address the blame cultures that still existed.'No-blame' and 'just' cultures were described to be needed, sometimes

MH1
Unsupportive reporting contexts "I don't think there is a culture of reporting near misses, and I doubt we vary much from the national situation."

AH2
"I can't think of any near misses that we have reported externally… I'm not sure how seriously they would take it even if you did…"

GP6
Assuming impact of learning from near misses "We would assume that organisations are learning effectively…" NB2 "No, no evidence, but I hope and I'm fairly certain that it has prevented future things."GP6 AH: acute hospital; MH: mental health organisation; AS: ambulance service; GP: general practice; NB: national/regional bodies.
with the conflation of the two.There was no consensus as to which culture was best, but all agreed that engaged and present leadership was required to develop a culture.
While participants described the importance of safety cultures in support of reporting and learning from PSNMs, they were also clear that other contextual factors needed addressing.Accessibility and usability issues with IRSs were reported.IRSs were the primary way to report all types of safety events across organisations.Other non-IRS routes for reporting, such as safety huddles, were thought to be useful.However, it was also heard that reports collected outside of the IRS rarely contributed to wider organisational learning as they were not collated.

Assuming the impact of learning from near misses
National participants described how they 'assumed' organisations were learning from PSNMs.Participants in organisations were less positive, and most were unable to provide examples of learning or actions following a PSNM.Any actions described focussed on increasing vigilance to safety risks, and in one case, process redesign.
No participant was able to evidence that learning from PSNMs had improved the safety of patients.Despite this lack of evidence, all assumed that learning from PSNMs had improved patient safety, and all agreed that attempting to learn from them was the right thing to do.Their justification was that the contributory factors to PSNMs were the same as to incidents and so, by addressing the causes of PSNMs, organisations 'must' be improving safety.

Improving learning from near misses
Participants described the need for learning from PSNMs to be seen as equally important as learning from harmful events.However, all acknowledged that impactful learning was more likely to follow harm.Participants also described other organisational needs to support improvements in learning from PSNMs.Themes to emerge included the need to: clarify what a PSNM is and why reporting is useful; develop appropriate safety cultures, and accessible and efficient IRSs; and develop guidance on analysing and effectively learning from PSNMs.

Discussion
Healthcare organisations have long focussed on reacting to and learning from harmful safety events. 20However, there is also learning potential in those safety events that have not reached a patient and/or not caused harm.While healthcare owes it to patients, families, and staff to understand why harm has occurred, all available learning opportunities should be explored to help improve patient safety.
Over 20 years have passed since the first healthcare reports highlighted the benefits of learning from PSNMs. 2,21The rhetoric to learn from PSNMs is oft repeated, but this study again found that learning is being lost. 5In contrast to the stated propositions at the beginning there is no standard definition of a PSNM used; PSNMs are rarely included in NHS processes; and there are assumptions rather than evidence that learning from PSNMs is improving patient safety.Potential learning from PSNMs is indeed falling into a 'black hole'. 22s to why there has been a lack of progress learning from PSNMs, this study highlights several systemic barriers to learningthere is variation in what is considered a PSNM; contexts are unsupportive of reporting safety events in general; and there is no national direction or encouragement to learn from PSNMs.][4]21 The authors had hoped to share examples of how organisations have been attempting to learn from PSNMs.The findings did not provide examples and no national guidance on how best to learn was found.If healthcare-based examples around how best to learn from near misses are lacking, then guidance needs to be sought elsewhere.As described, some industries have found value in learning from near misses. 10,11The following discussion considers the needs raised by study participants for the improvement of learning from PSNMs, with insights taken from other industries.

Answering the fundamental questions
Participants wanted clarification as to what is a PSNM and why learning from them is useful.][25] Variation in the interpretation of the term 'near miss' demonstrates a lack of agreement as to the key features of a PSNM.This is unsurprising considering the differences in guidance provided internationally.While the WHO has defined the term, 6 bodies such as the United States' National Coordinating Council for Medication Error Reporting and Prevention, and the United Kingdom's NHS, avoid using the term, instead referring to errors, prevented incidents, and no harm. 1,26In general, there has been a move away from referring to near misses, with increasing reference to 'good catches'. 27A change in terminology may foster a more positive view of PSNMs, but research in other industries shows the dangers of seeing them as successes. 28larifying the features and definition of a PSNM is beyond the ability of this study, but the authors have reflected on this elsewhere. 25In support of clarifying the features, this study found importance in the presence of interruptions that prevent events from progressing.Interruptions align with how near misses are viewed in other industries where they consider the role of 'controls'. 29,30The interruptions referred to by participants are the measures in place (controls) to prevent hazards from turning into incidents. 30The effectiveness of a control will depend on its type, when, and how it acts.In the PSNMs discussed in this study, controls commonly relied on staff vigilance, such as spotting that a patient had been prescribed a contraindicated medication before administration.Controls that require a human to act, particularly without planning, represent vulnerabilities and more effective controls are needed.A more effective control may be an electronic prescribing system with forcing functions to prevent the prescribing of a contraindicated medication.
With regards to why learning from PSNMs is valuable, the finding in relation to the role of interruptions/controls is pertinent.Interpretation of a PSNM with respect to controls draws attention to the resilience of systems. 11A PSNM can therefore provide insights into the contributory factors to events, and the presence and effectiveness of controls to prevent harm.This also means that conflating PSNMs and other safety events may hinder learning by drawing attention only to the contributory factors.

Creating the conditions
It is well known that safety cultures influence the reporting of and response to safety events. 31,32This study again found safety cultures to be important in learning from PSNMs.Alongside culture, the design and accessibility of IRSs were found to be a barrier to reporting PSNMs.
Creating the conditions to support learning from PSNMs includes the need to address the sociotechnical barriers to their reporting.In modern healthcare, reporting needs to be easy and quick.Even better would be to remove the need for staff to identify and report PSNMs with automated processes; these are being increasingly explored. 33As for the design of IRSs, accessibility and ease of use need to be balanced with the collection of useful information.Industry literature demonstrates the need for IRSs to collect information about how systems contribute to events, 34 and what controls are present and how they perform. 35eturning to culture, the conflation of just and no-blame cultures was found in this study.In industry, just rather than no-blame are advocated for. 36Just cultures maintain professional accountability. 37Just is also more feasible than no-blame cultures. 38To develop the right culture, this study echoes two important factors that have a role 39,40 -(1) visible leaders who uphold the culture, and (2) longstanding teams with psychological safety.

Effectively learning from near misses
No guidance for PSNM learning was found during this study.Participants wanted practical guidance to support organisations to learn in ways that contribute to system improvements, rather than just increasing vigilance.It was also acknowledged that, even with the best will, PSNMs are less likely to stimulate action and improvement as no harm occurs; PSNMs alone may not provide impetus for change. 41n industries such as aviation, near misses are not considered in isolation.They are a form of safety intelligence contributing towards proactive hazard identification. 42,43While an industry may investigate a 'significant' near miss (i.e. one that poses a significant risk of future harm), information about all near misses are aggregated with other intelligence.That aggregation can highlight themes in hazards, contributory factors, and the effectiveness of controls, [44][45][46] and builds an evidence base in support of improvement actions.Taxonomies are used to categorise the information from intelligence to support aggregation. 470,23,27,32 Healthcare has been criticised for not being more proactive with safety management and is limited in its use of different intelligence sources. 48Healthcare taxonomies do exist for the categorisation of intelligence, 6,26,49,50 but this study did not see these applied to PSNMs, let alone other forms of intelligence.The limited learning seen from PSNMs may be, in part, due to a lack of aggregation with other intelligence to identify learning themes.Also, while healthcare taxonomies exist, they are limited by their lack of consideration of the presence and effectiveness of controls 35 ; further research is required.

Strengths and limitations
There has been no examination of the state of PSNM learning in the NHS since the early 2000s. 2 This study was therefore appropriate and had the potential to benefit patient safety.The methodology provided a rich understanding of PSNMs in the sectors chosen and theme saturation was reached.The study was limited in that it did not consider all NHS sectors.Also, because of the PSNM definition variation, the ability to compare processes and outcomes was challenging. 51It is also acknowledged that the findings around a lack of impact of learning from PSNMs will be, in part, due to the lack of process for and underreporting of PSNMs.However, organisations did not know how to learn from PSNMs even if they were reported; this study, therefore, adds value.

Conclusion
PSNMs are plentiful and can offer insights into the resilience of healthcare systems to support improvements in patient safety.However, their learning potential is not being exploited due to the lack of an agreed and standard definition, unsupportive reporting contexts, and no encouragement.As a result, there is an absence of a process for learning from PSNMs in the English NHS.How best to develop learning processes requires further research, but this study offers insights in support of defining, reporting, and investigating PSNMs.The findings also demonstrate the need for agreement as to the features of a PSNM before a definition can be standardised.

Figure 1 .
Figure 1.Thematic map of findings from the qualitative case study.

Table 1 .
Example quotes from participants in support of the themes.ThemeQuote Source Variations in safety event schema "We have tried to move away from the term near miss and would now refer to these as no-harm incidents…"NB2No specific processes for near misses "…they join the list of overdue incidents for closure because our attention is drawn to [harm]…"