Background to the problem: how do we learn from when things go wrong in healthcare?

Patient safety is a prominent focus for healthcare providers, regulators and commissioners. Since the seminar publication of To err is human, the role of errors in healthcare was pushed into the public eye—for example, with the National Academy of Science estimating in this publication, in the late 1990s, that between 44 000 and 120 000 Americans die due to ‘people making mistakes’.9 While between the late 20th century to the current day, there has been significant academic contributions to argue healthcare incidents are greater than ‘mistakes’, the focus on ‘people making mistakes’ has been hard to remove from the discourse of patient safety. Reason’s principles of error management, and the famous metaphor of the ‘Swiss cheese’ that ensued, argued that errors were symptomatic of latent failures embedded in systems of work.10 Reason argued that when these latent failures, lying dormant, aligned, catastrophic errors could follow and our knee-jerk reactions to these catastrophes could result on us focusing on the individuals at the ‘sharp end’ of the system.11 However, currently, insight into patient safety incident management still identifies that blame of individuals is commonplace in the response to errors despite the movement towards ‘systems thinking’.12 13

Systems thinking, in its simplest, is appreciating both the explicit and tacit processes that surround a system of work. Embracing systems thinking is almost synonymous with embracing the complexity of healthcare and appreciating that incidents in healthcare may not follow a linear causation process.14 To fully appreciate the system in healthcare incident investigation demands a system-focused investigatory method and it has long been argued that the prominent approach to investigation in healthcare, the root cause analysis (RCA), does not fully allow for a systems focus.5 This is because, in part, the RCA method argues for incidents to be reduced to the key ‘root’ of the problem or context. The sharp end of a failing system may relate to an individual omission—for example, failing to prescribe or administer a drug—and therefore, the RCA investigatory method can result in person-focus findings and person-centred recommendations. However, this individual omission is likely symptomatic of a much more complex problem.15 Indeed, in a systematic review conducted by Martin-Delgado et al, it was argued that while RCA was a useful tool to understand a specific incident, it was less useful in implementing improvement and recommendations that would prevent their recurrence.16 Investigations that are based on systems-focused methodologies or frameworks, for example, Accimaps and System Engineering Initative for Patient Safety (SEIPS),17 18 have been argued to result in actions or recommendations that are more likely to result in sustained improvements to the overall system of safety.19

Problem description

Dekker and Hollnagel identify that for every ‘accident’, there could be 10 incidents, 30 near misses and 600 unsafe acts associated with that 1 accident.20 In healthcare, this accident would likely to be an incident causing death or severe harm, and incident referring to incidents reported with low or no harm to a patient. For example, for an administration error, there may be multiple times where the wrong drug could have been administered, but a safety barrier alerted staff to this, and it was logged as a near miss. Similarly, there may have been hundreds of times where the second checking of certain processes was missing, lax, performed under suboptimal conditions, or done while performing other tasks. In the latter two cases, these near-misses and unsafe acts may not have caused harm, and therefore, may not have alerted the organisation to investigate, compared with the one case where harm occurred to the patient. While there has been acceptance that near-misses and low harm incidents are as important as incidents resulting in harm,21 practice across healthcare has seen more resource being earmarked for investigating incidents that result in harm. For example, from 2015 to July 2022, the National Health Service (NHS) was contractually obliged to investigate ‘serious incidents’ (categorised by harm to the patient) via an RCA-style investigatory method22 before adopting a more systems-focused policy from August 2022.23

While we have seen a movement to adopt system-focused methods for investigating incidents, healthcare organisations still have little support in templates or guidance on how they can group together incidents and focus on the system of safety via a triangulation of near misses, low harm incidents and adverse events resulting in harm. Systems of safety are complex, and therefore, triangulating, and aggregating incidents pertaining to the same system of safety can provide an insight into the parts of the system that are failing and why. If incidents pertaining to the same system, all with individual insights with varying impacts on patient care, are aggregated, then we would be able to appreciate in what context a part of a system or process fails and how this affects other interconnecting parts of the system. This is almost akin to providing an insight into a stress test of a system—with the focus on how resilient the components of the system are in different contexts. This paper argues that a focus on systems is only part of the solution to understanding safety in healthcare, and the missing piece in this quandary is the need to group together multiple incidents in a systematic and system-focused way. The next section will present the themed review tool and then go on to argue how its use in a real-world example presents and encourages system-focused, valid and specific recommendations. The example used in this case was conducted in a multidisciplinary team with varying levels of knowledge of safety science, working in an acute healthcare setting.

Intervention: a themed review template for use in healthcare

Design and format

The themed review template can be found as online supplemental material to this paper. The template is set up into five stages: (1) description of the reference cases, (2) description of the safety system, (3) classification of contributory factors for reference cases, (4) narrative analysis and (5) safety recommendations based on the thematic review. Each stage is narrated to include instructions to the investigator on how they complete the individual steps. Stage 1 refers to noting down the reference cases to note key descriptions for each incident, for example, harm occurred to patient or date of incident. Stage 2 refers to the description of the safety system before any theming of the incidents together. To describe the safety system the template guides the investigator to process map the system of safety, with two examples. The mapping of systems is important in understanding better how a system of safety is set up for areas of care—for example, falls management—and this stage provides a critical eye on what the system of safety is in reality. This refers to Shorrock’s work on varieties of human work24—where there may be a dichotomy between work as prescribed (ie, the designed system of safety) and work as done (ie, the reality of work on a day-to-day basis). This stage asks for a mapping of the system into compartments and then guides the investigator in filling in an ‘after action review’ style format to identify how, across the reference cases, these stages are failing. The aim of this stage is to critique the stages of a system that may be overly relied on but not actually be contributing to overall safety—referred to as ‘safety clutter’.25 An example of this stage can be seen below in figure 1 with the example of management of the risk of falls. Each box will have an in-depth review of how this aspect of the system was engaged with for each reference case which then paves the way for stage 3, the classification of contributory factors for each reference case.

Supplemental material

[bmjoq-2022-002020supp001.pdf]

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Example of stage 2 and mapping a system of safety.

Stage 3 aims to help investigators categorise factors that affected each reference case. The importance of this stage is to ensure that the investigator(s) are encouraged to think wider than the individuals involved in the reference cases when thinking about where things went wrong in a safety event. To do this, it was important that a framework could be selected to allow for this which would aid the initial thematic analysis of multiple safety incidents. Unlike the research sphere where Braun and Clarke’s work on thematic analysis paved the way for sound methodological analysis of narrative data,19 currently little guidance exists for practitioners to conduct sound thematic reviews on narrative data surrounding safety incidents. While Braun and Clarke’s six steps for thematic analysis has been shown to provide a robust and thorough methodology, there is a pragmatic concern that this methodology of analysis and theming is not appropriate for use in this context. For example, narrative data for each safety incident may be incomplete and of differing standard when compared and grouped. This is unlike thematic analysis methods which use the same data set which has been guided and generated by the same epistemology, methodology, and methods. However, there are lessons we can learn from thematic analysis and apply to this context.

First, healthcare incidents and data collected from incident investigations have large elements of description to them—for example, categorisation of where the incident happened, the harm level, the part of the system that failed (eg, traditional ‘root causes’), and foci for improvement, and therefore, themed reviews of multiple safety incidents may benefit from deductive analysis based on set criteria.26 Following this deductive approach for the development of the themed review template, it was necessary to find a universal code or framework that allows the investigator to categorise cross-cutting themes that are pertinent to a particular aspect of safety. In practical terms, this meant identifying a framework that would be applicable to themed review of multiple safety incidents portfolios, for example, medication errors, diagnostic errors, management of a septic patient. The final framework chosen was the contributory and mitigations factors classification27 although other frameworks considered included SIEPS18 and The Yorkshire Contributory Factors framework.28 The contributory and mitigations factors classification has similar benefits to the aforementioned considered approaches in that it provides different categories of factors affecting safety at different ‘levels’ in the system. However, its main benefit and reason for inclusion is its clear and simple components within each ‘level’ of the system. These three levels relate to (1) organisational factors, (2) task/environment factors and (3) individual factors and each level has three subgroups and explicit components. An example of a subgroup of (2) task/environment factors can be seen below in figure 2—a subgroup referring to task factors.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Example of contributory factors.

While the use of such a deductive framework could be argued to hinder generation of contributory factors from an inductive perspective, the mitigation to this in the template was to include options for free text in stage 4. The template guides investigators to mark a X next to each reference case and criteria where this component affected the reference case.

Stage 4 is a more narrative and interpretive step where multiple reference cases are grouped together where the same contributory factor has been identified in stage 3. The aim of this stage is to encourage the investigator(s) to consider the commonalities and differences in the way the contributory factor affected each case. For illustration, in the example of figure 2 above and the task factor ‘procedural or task design and clarity”, this may be noted multiple times as affecting multiple reference cases and incidents. Through a hypothetical example, this task clarity could relate to a themed review of mismanagement of septic patients and non-adherence to a policy like the sepsis bundle. Grouping together incidents where the policy, or task, was not adhered to allows us in this stage to consider many question—for example (1) what was the context around the patient’s deterioration? (2) what was the staffing on that shift and how did it affect ability to carry out task? (3) what were the reasons the task was not carried out as anticipated? (4) are these variations commonplace and this is the first time we have been alerted to them via an incident? Allowing for this interpretation is encouraging an appreciation of complexity within a system—that is, different parts of a system will interconnect in spontaneous ways depending on the context (eg, staffing/acuity of patients/staff experience) and grouping together a group of incidents can allow us an insight into the work as done and challenge assumptions around work as imagined. An example of this reflection could be a themed review of falls which identifies that in all incidents, the risk assessment was completed to the standards expected. Theming together these cases may encourage investigator(s) to then critically consider how risk assessments affect the safety system, and therefore, how reliable reliance on these measures are.

Stage 5 is the last stage of the themed review which refers to the safety recommendations following the themed review. These recommendations are adapted from Shand et al’s work29 where these three recommendations refer to the improvements categorised into ‘fixes’, ‘improvements’ and ‘changes’. This useful framework for recommendations allows investigator(s) to appreciate the complexity and challenges that come with improvement to systems of safety. ‘Further insight’ has been added to the categories of recommendations which identifies the possibility that the themed review has resulted in more questions than answers relating to a safety issue—and therefore, a recommendation could be to collect further safety insights for example, gathering staff insights into the system of safety via surveys or focus groups.

Example of use and impact on safety recommendations

This next section will identify the impact the themed review can have on understanding complex systems of safety and how recommendations can be more system-focused than person-focused. The need for recommendations to be system-focused is due to research identifying that person-focused recommendations are less likely to be successful in implementing change. The hierarchy of intervention effectiveness30 is a useful framework when critically reviewing the recommendations that emerge after incident investigations. Through this framework, we can identify that reminding staff or creating checklists, defined as ‘people-focused’, are argued to be less effective than forcing functions or standardisation of processes, here defined as ‘systems-focused’. This is in part the argument for more systems-based investigations because they move investigators away from recommending person-focused solutions. However, it is argued that recommendations after investigations focus more on person-based solutions—for example, in research conducted by Woods et al analysing safety recommendations, 37% of all recommendations were person-focused, compared with 1% of recommendations focused on forcing functions—that is, changing the system to design the error out of the system.31 This in part may relate to the difficulty in being able to recommend solutions that are system focused.

Approaching safety issues via a thematic review approach may allow for a focus on the system rather than individuals. This is in part due to the fact that if incidents reoccur, in different contexts (eg, shift patterns or different professionals), then it is evident that an issue transcends individual accountability. The worked template presented in this paper illustrates how multiple incidents can be grouped together and analysed using a system-based classification tool. Shifting local and organisational norms away from individuals fallibility or ‘human error’ to systems thinking and ‘system failures’ is complex and difficult, and this approach is argued to encourage investigators of safety incidents to consider safety concerns through the lens of multiple incidents and failings, rather than one.

The themed review template, and its focus on multiple incidents and the contributory factors classification framework, allowed for a systems focus. As well as providing a worked example of a thematic review of six cases of deteriorating patient, it also compares findings for these six cases which were individually investigated. This was chosen as a method to identify if the thematic review approach allowed for a wider, systems, focus compared with all six incidents which were investigated in isolation of each other.

After noting the cases for inclusion in the review, the system of safety for management of deteriorating patient was mapped, which can be seen in figure 3. This system of safety relates to the system barriers that exist to safeguard patients from mismanaged deterioration. Again, having a focus on system barriers early on in this template allowed for an initial appreciation of the role of system barriers or controls, rather that person-focused barriers or controls.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

mapping a system of safety for deteriorating patients.

Then, each safety barrier was assessed using the reference cases. This allowed for investigators to assess the strength of all these barriers. For example, with safety barrier 2 (senior input into management of patient), it was identified that in all the six reference cases, there was a delay or omission in senior review. On top of this, the assessment of this safety barrier identified that there were different reasons why this senior review was omitted or delayed. These reasons related to lack of clear communication of patient status, differing mode of assessment, different data used to assess patient from senior medical team and conflicting information on what type of senior review was needed. At the first stage of this review, the complexity of this safety barrier is clear to see. For illustrative purposes, one of the recommendations from the original RCA was to ‘remind senior doctors to review patients who are at risk of deterioration’. This is a clear example of how individual investigations versus a themed review of multiple investigations differs and how a themed review allows space for exploring the myriad of different reasons why senior review may not have been possible for deteriorating patients.

Figure 4 is an extract from stage 3 of the themed review and identifies a worked example of assessing the contributory factors for the six included reference cases. This section identifies that staff supervision was a particular factor for four of the incidents.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Assessing contributory factors to the reference cases.

Using the operational management factors domain as a continued example, stage 4 identifies a narrative review and an example of this in practice for this case can be seen in figure 5. This narrative review asks for the investigator to assess the similarities, differences and contextual factors surrounding the reference cases. The nuances and similarities in the issue of senior review are explored, and this stage identified that the omission of a senior review was twofold—the lack of review in its entirety and the review being incomplete or not in person. This latter finding is useful as it identifies the work as done—where senior review may be completed over the telephone and in some cases, this is a patient safety risk due to inadequate holistic assessment of the patient.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

A breakdown of the narrative analysis step.

For ease, so far this section has followed one factor identified in the themed review through all the stages of the review. In its entirety there were five key factors identified which shaped the recommendations in stage 5. A comparison of the safety recommendations from those from the themed review, versus the individual RCA investigations identify the focus the themed review placed on assessment and improvement of the system of safety. table 1 below identifies a comparison of recommendations for the themed review versus individual RCA investigations.

The focus of this paper is not to formally evaluate if the template has an impact on the safety outcomes related to management of deteriorating patients but to identify if theming together incidents can facilitate systems-thinking and complexity in patient safety. Table 1 identifies that the findings and recommendations from the thematic review were more systems-focused than individual incident investigation. Comparing this to the RCAs done for individual cases, the recommendations are person-focused and relate to reminding staff to follow policy—in this conduct observations at the correct time and review patients in the correct way—without understanding why there may be a dichotomy between the work as done versus the work as prescribed or imagined.

Limitations and focus for the future

With a movement towards systems-thinking in patient safety, thematic reviews as an investigatory approach provide an opportunity to look at a cluster of incidents through a system-focused lens. The purpose of this paper is to identify the merit of the investigatory approach and provide an example of a template used. A key limitation of this paper is the lack of usability assessment, which is an important facet of whether the tool works in practice. This was beyond the scope of this study; however, future research should look to ascertain what level of safety science or patient safety education is needed to engage with the concepts fully.

Another important consideration for the success of thematic reviews is the quality of investigation which is used as the data for the thematic review. For example, RCAs have been argued to reduce complex problems to simplistic or person-focused actions.5 Therefore, the quality of the investigation feeding the thematic review may inhibit the ability to ascertain system-based findings. This could be mitigated by keeping a constant skill level (eg, training skill) or investigating all the cases with the same investigators with the requisite investigatory knowledge.

It is also important to remain pragmatic about the difficulties in improving aspects of patient safety following investigations. This paper argues that thematic reviews, as an investigatory method, provide a better understanding of the system of safety around safety concerns. However, the link to better insight and improved safety may be intuitive but certainly not linear. Therefore, it is also essential that future research into thematic reviews as an investigatory method focus on the impact of increased insight on more effective actions and recommendations and ultimately, safety outcome measures. The potential success of thematic reviews centres around better insight into a system of safety resulting in an ability to facilitate targeted improvement. However, evidence of this relationship is lacking, and therefore, this theoretical assumption should be tested across a range of different safety topics. A comparison between recommendations from thematic reviews versus individual investigations would be an important research focus and could identify if recommendations from themed reviews were more sustainable or effective over time.