Gaps, traps, bridges and props: a mixed-methods study of resilience in the medicines management system for patients with heart failure at hospital discharge

Introduction Poor medicines management places patients at risk, particularly during care transitions. For patients with heart failure (HF), optimal medicines management is crucial to control symptoms and prevent hospital readmission. This study explored the concept of resilience using HF as an example condition to understand how the system compensates for known and unknown weaknesses. Methods We explored resilience using a mixed-methods approach in four healthcare economies in the north of England. Data from hospital site observations, healthcare staff and patient interviews, and documentary analysis were collected between June 2016 and March 2017. Data were synthesised and analysed using framework analysis. Results Interviews were conducted with 45 healthcare professionals, with 20 patients at three time points and 189 hours of observation were undertaken. We identified four primary inter-related themes concerning organisational resilience. These were named as gaps, traps, bridges and props. Gaps were discontinuities in processes that had the potential to result in poorly optimised medicines. Traps were features of the system that could produce errors or unintended adverse medication events. Bridges were features of the medicines management system that promoted safety and continuity which ensured that, despite varying conditions, care could be delivered successfully. Props were informal, temporary or impromptu actions taken by patients or healthcare staff to avoid potential adverse events. Conclusion The numerous opportunities for HF patient safety to be compromised and for suboptimal medicines management during this common care transition are mitigated by system resilience. Cross-organisational bridges and temporary fixes or ‘props’ put in place by patients and carers, healthcare teams and organisations are critical for safe and optimal care to be delivered in the face of continued system pressures.

Furthermore, it should be emphasized that the study applies to a specific patient population, ie those suffering from systolic decompensation, that represent the minority of patients subject to readmissions due to heart failure. Most re-admissions due to decompensation take place in patients with preserved ejection fraction (pEF) and are often related to comorbidities. (Roger VL. Epidemiology of heart failure. Circ Res. 2013;113:646-59. doi:10.1161/CIRCRESAHA.113.300268.) Since RCTs performed up to now have not been able to identify treatments with a prognostic impact in patients with heart failure and pEF, in this specific group therapeutic uncertainty is greater than in patients with heart failure and reduced EF; since there are only a few strong recommendations in therapeutic guidelines about heart failure with pEF, the prescriptive variability is expected to be greater in this field, as greater is the probability of errors and/or changes in the management of medicines through the transition from hospital to primary care. Given that the presence of comorbidities further complicates the management of medical therapy, it would be interesting to know the distribution of comorbidities in the patients included in this study.
With regard to bridges and props, the added-value of the titration clinics managed by pharmacists can be reinforced by the recent demonstration of the better arterial pressure control in hypertensive black people through pharmacological intervention led by pharmacists in barber shops versus the usual care. (N Engl J Med 2018;378:1291-1301. DOI: 10.1056/NEJMoa1717250).
In short, this study analyzes a delicate aspect of medical practice, namely the management of medical therapy at discharge from hospital. This is a complex and poly-parametric field, that has been correctly analyzed by the authors from multiple perspectives (hospital doctors, GPs, patients, pharmacists) and with an appropriate methodology. The study sheds light on some virtuous aspects of drug management at hospital discharge. In this sense it can be a basis (and/or an integration) for methodologically sound clinical studies (eg cluster RCTs). However, just because it is a case study, I consider mandatory to replace the term "demonstrates" with the term "suggests" (third line of the discussion). In short, I would accept the paper with minor revision.
I have a few suggestions for improvement, which are more around the definition and use of Safety-II concepts rather than specifically on the topic of medicines management.
(1) terminology: it would be helpful to clarify key terms and concepts. For example p3/l42 "errors" -are these human errors, medication errors, or what kind of errors? p4/l3 "specific risks" and p15/l31 "minimise the risk of error" -again, how is risk defined? What does minimising the risk of error mean? Reducing the likelihood of its occurrence?
(2) Safety-II: how do the concepts of error and risk sit with Safety-II? Error hasn't been defined in the paper, but is usually something like deviation from a procedure / standard, or a negative outcome. Risk could be defined as the combination of likelihood of occurrence of a hazard and the severity of the consequences. These are all negative. Safety-II does not normally operate with such concepts, so it would be helpful to include the authors' interpretation.
(3) Novelty: I believe the novelty of the findings (in terms of Safety-II) might be slightly overstated. I would suggest consulting this now pretty much seminal reference (18 years old and still a great read): Cook, R.I., Render, M. and Woods, D.D., 2000. Gaps in the continuity of care and progress on patient safety. Bmj, 320(7237), pp.791-794. Since, there have been numerous publications that essentially cover the same ground of gaps, traps etc albeit with different vocabulary. The 3 books (soon 5) on Resilient Health Care might also be worth consulting, as well as the special issue on Resilience Engineering published in Reliability Engineering & System Safety (2015).
(4) Learning: p15/l49 "Learn from failure" -I would argue this is a misrepresentation. Safety-II suggests organisations should learn from what goes right as well as from what goes wrong, with a focus on everyday clinical work. "Learn from failure" could be better called "Learn from experience". If you are interested in this specific topic, you might find this article of some use: Sujan, M.A., Huang, H. and Braithwaite, J., 2017. Learning from incidents in health care: Critique from a Safety-II perspective. Safety Science, 99, pp.115-121.
(5) Accimap -this is somewhat sprung upon the reader. It should either be explained or not mentioned at all.
(6) ResiMap -would require further clarification. How is this different from FRAM? What's the actual theoretical underpinning?

REVIEWER
Dr Duncan McNab NHS Education for Scotland, United Kingdom REVIEW RETURNED 15-May-2018

GENERAL COMMENTS
Overall I think this is a really useful way to convey the principles of system resilience to frontline teams. Below are some specific comments for consideration and some general comments that I think would strengthen the discussion section. I think the discussion could include a fuller comparison with the existing resilience literature.

INTRODUCTION
Page 3 line 34 -does it follow that medicines need to be optimised -need another line re control by medicines but can cause harm??

METHODS
Page 4 line 27 -4 health economies -could this be clarified? There are 4 sites but 5 wards and one HF clinic -I think that this needs clarified -also why were these sites chosen -purposively sampled? Pager 3 line 52 There is a more up to date definition of resilience In the fourth book (Resilience Engineering in Practice, 2010) the definition is given as, "The intrinsic ability of a system to adjust its functioning prior to, during, or following changes and disturbances, so that it can sustain required operations under both expected and unexpected conditions." This is the given on the website of the resilient healthcare network is, "A system is resilient if it can adjust its functioning prior to, during, or following events (changes, disturbances, and opportunities), and thereby sustain required operations under both expected and unexpected conditions." Page 4 line 18 -Resilience is therefore more than compensating for weaknesses but also responding to opportunities. I think you have studied this -for example by considering opportunities for patients to contribute to creating safety. Page 4 line 21 typo "where resilience in the exists" Page 5 line 6 -"A quota sample of between 16-24 admitted patients was constructed to allow for attrition, aiming for 16 complete datasets." For me this could be clearer. It could also state this is the composite total between the four sites. Page 5 line 38 -how were healthcare staff recruitedconvenience sample/ purposive?? Page 6 line 6 -document analysis on page 4 line 30 stated that using case note review for document analysis, This is not mentioned here in the analysis section. Page 6 line 10 -states "Examples of system resilience at care transitions and risks in the system were extracted using a framework that mapped them according to the point in the transition to which they related and to the resilience element (or lack of) they evidenced." Are these examples of resilience potential recorded in official documents? Or does this refer to patient case note review and examples of resilience found? I don't think you can say a document (protocol/guidance etc) shows system is resilientmaybe it indicates that there is potential to be resilient. Page 6 line 24. It states that data analysis was iterative then that "The research team met several times to discuss the data synthesis and analysis method and the emerging themes." I think this should be more specific -did the research team meet between data collections to discuss the data synthesis and analysis method and the emerging themes? Page 6 line 31 -The emerging analysis was thematic. Could this be clarified -my understanding of framework analysis is that thematic analysis is conducted as part of the process -so themes emerge not the analysis? Page 8 line 15 second time heading "Results" has appeared Page 8 line 30 -a 'gap' defined as a discontinuity of key process. The following is given as an example of a gap: "For the latter, we identified no standardised processes for informing patients about their medicines and, while hospital policies stipulated that patients should be informed, and gave details of the types of information patients should have, there was no guidance on optimal methods for informing patients about their medicines or training in doing so." The 'gap' is that the patients did not receive the correct information. The lack of guidance is a contributory factor (probably one of many) in why this happens. The lack of guidance and training is certainly a gap in the 'system-as-found' but is it a discontinuity of a key process or does it just not exist? It may be that this actually is a trap as it relates to the way the system is designed -there is no guidance or training. On page 9 line 34you infer that a similar problem in primary care is a trap and not a gap.
Page 10 line 40 -GP staff says prioritise based on risk readmission -is this formal and training given? If so, then agree it is a bridge, if not it is a prop. Props -informal resilient behaviour -but as soon as this becomes the formal system then do they become bridges?

TABLES
In tables -few abbreviations need expanded HCA and MCCA HFSN TTO Table 7 row 8 -What are ambulatory services? (1) Terminology: it would be helpful to clarify key terms and concepts. For example p3/l42 "errors" -are these human errors, medication errors, or what kind of errors? p4/l3 "specific risks" and p15/l31 "minimise the risk of error" -again, how is risk defined? What does minimising the risk of error mean? Reducing the likelihood of its occurrence?
We have addressed this point We defined 'traps' as features of the way the medicines management system was designed or managed that might produce medication errors defined as a 'failure in the treatment process that leads to, or has the potential to lead to, harm to the patient' (Ferner & Aronson 2006) Sometimes the props, were put in place despite organisational pressure, for example to discharge patients and free beds..
(2) Safety-II: how do the concepts of error and risk sit with Safety-II? Error hasn't been defined in the paper, but is usually something like deviation from a procedure / standard, or a negative outcome. Risk could be defined as the combination of likelihood of occurrence of a hazard and the We appreciate that we have not been clear on this matter and have improved the text accordingly.
We have made multiple text changes to enhance clarity severity of the consequences. These are all negative. Safety-II does not normally operate with such concepts, so it would be helpful to include the authors' interpretation.
We believe that the language of Safety 1 can be used to understand where there are opportunities to improve system performance through understanding where variations in performance have produced less than optimal outcomes.
In a recent article Rebecca Lawton points out that the system approach now characterised as safety 1 described in 'An organisation with a memory' encompasses both learning and resilience. Pager 3 line 52 There is a more up to date definition of resilience In the fourth book (Resilience Engineering in Practice, 2010) the definition is given as, "The intrinsic ability of a system to adjust its functioning prior to, during, or following changes and disturbances, so that it can sustain required operations under both expected and unexpected conditions." This is the given on the website of the resilient healthcare network is, "A system is resilient if it can adjust its functioning prior to, during, or following events (changes, disturbances, and opportunities), and thereby sustain required operations under both expected and unexpected conditions."

Definition updated and references added
This in turn promotes a more dynamic attitude to performance through resilience which we define here as the ability for a system and the individuals therein to adjust prior to, during of following bounce back after any changes disruption or failure or disturbances or in the face of ongoing, sustained pressure Page 4 line 18 -Resilience is therefore more than compensating for weaknesses but also responding to opportunities. I think you have studied this -for example by considering opportunities for patients to contribute to creating safety.
More specifically, the study was designed to understand how the system compensates for weaknesses and maximises opportunities in order to deliver safe yet optimal treatment.
Page 4 line 21 typo "where resilience in the exists" We have corrected this typo Page 5 line 6 -"A quota sample of between 16-24 admitted patients was constructed to allow for attrition, aiming for 16 complete datasets." For me this could be clearer. It could also state this is the composite total between the four sites.
A quota sample of 4-6 patients in each site was constructed, aiming for at least 16 complete datasets in total in the four areas Page 5 line 38 -how were healthcare staff recruited -convenience sample/ purposive??
Have reworded this sentence A range of healthcare professionals involved in medicines management were selected following ward observations Page 6 line 6 -document analysis on page 4 line 30 stated that using case note review case notes and communications such as for document analysis, This is not mentioned here in the analysis section.
discharge letters. We aimed to include a range of healthcare professionals involved in medicines management.
Page 6 line 10 -states "Examples of system resilience at care transitions and risks in the system were extracted using a framework that mapped them according to the point in the transition to which they related and to the resilience element (or lack of) they evidenced." Are these examples of resilience potential recorded in official documents? Or does this refer to patient case note review and examples of resilience found? I don't think you can say a document (protocol/guidance etc) shows system is resilient -maybe it indicates that there is potential to be resilient.
They were identified as potential sources of resilience Examples of potential system resilience at care transitions and risks in the system were identified and Page 6 line 24. It states that data analysis was iterative then that "The research team met several times to discuss the data synthesis and analysis method and the emerging themes." I think this should be more specific -did the research team meet between data collections to discuss the data synthesis and analysis method and the emerging themes?
We have clarified this The research team met several times both during and following data collection to discuss the data synthesis and analysis method and the emerging themes Page 6 line 31 -The emerging analysis was thematic. Could this be clarified -my understanding of framework analysis is that thematic analysis is conducted as part of the process -so themes emerge not the analysis?
We have deleted this confusing line Page 8 line 15 second time heading "Results" has appeared Thank you we have deleted this Page 8 line 30 -a 'gap' defined as a discontinuity of key process. The following is given as an example of a gap: "For the latter, we identified no standardised processes for informing patients about their medicines and, while hospital policies stipulated that patients should be informed, and gave details of the types of information patients should have, there was no guidance on optimal methods for informing patients about their medicines or training in doing so." The 'gap' is that the patients did not receive the correct information. The lack of guidance is a contributory factor (probably one of many) in why this happens. The lack of guidance and training is certainly a gap in the 'system-as-found' but is it a discontinuity of a key process or does it just not exist? It may be that this actually is a trap as it relates to the way the system is A better knowledge of gaps allows staff to anticipate where problems may occur and take action to avoid them. Props in the system are indicators of how flexible staff and teams are and healthcare systems can learn from the temporary fixes put in place and knowing where bridges have successfully joined up care can help systems learn and be better placed to innovate elsewhere.

Work as imagined versus work as done
An enhanced view of the system using this lens allows policy-makers to understand the gap between work as imagined versus work as done and better understand how policies and guidelines are actually implemented (or not) in healthcare organisations by staff who adjust their performance to deliver care in a complex system. This may serce to closer align work as imagined versus work as done.
Page 16 line 5 -policymakers should understand resilient action -perhaps, but not to this level -this is very important for local teams to learn from. Understanding system gaps, traps, bridges and props and sharing this knowledge. Policymakers should appreciate the actions of individuals and teams to provide safe care despite conditions.
Policymakers should recognise the attempts made routinely by healthcare professionals and healthcare teams to learn from their clinical experience and apply this learning to increase system resilience by delivering safer care for patients despite disruptive conditions, such as disconnected communication systems, varying staffing levels and the under-provision of formal training, for example in discharge and care transfers.
Page 17 line 30 -the Functional Resonance Analysis Method is surely worth mentioning as it comes from the Safety-II and Resilience Engineering world and can help show relationships between functions within a system that may help demonstrate functions that bridge or prop gaps. I'm not sure proposing ResiMap is helpful as there is no detail of what it is. For me most of early resilience research focuses on props and it is great that some See section "Implications for future research" research into bridges/props show system level change to improve resilience Conclusions Not sure you need this sentence in the conclusion: For example, some GP surgeries have systems in place to ensure the timely and efficient processing of discharge information We have removed this sentence

REVIEWER
Mark Sujan University of Warwick, UK REVIEW RETURNED 08-Jul-2018

GENERAL COMMENTS
The authors have addressed previous suggestions, and I believe the manuscript has improved.
The paper makes a useful contribution towards highlighting the importance of studying everyday clinical work in order to better understand how healthcare organisations provide good quality care, and how they can improve further. A strength of the study is the large sample size (for a qualitative study), and the inclusion of a significant number of patient interviews.