Solutions to wicked problems

As we move forward, we would like the balance between studies to shift from those focusing on describing problems to those focusing on solutions. We highlighted this in a recent editorial noting that we are not interested in describing ‘safety problem Y in every country of the world or in every different setting’.8 Descriptive studies of safety problems done in a single country or single institution will only be considered if they move the field forward by, for example, analysing how context influences the causes or prevalence of different safety issues or by gaining more insights into the underlying mechanism(s) through which safety problems might occur. Descriptive studies may also be considered if they describe a new or poorly understood safety problem, especially if it is common or has a large impact.

Equity is a good example of an area for which it is time to concentrate on solutions. We know that inequity in access to high-quality healthcare is a ‘wicked’, seemingly intractable problem across settings. We also need to be careful not to ‘jump to solutions’ that may seem logical but end up exacerbating inequities. Occasionally, promising solutions emerge such as the use of audit and feedback to increase physicians’ rates of identifying and responding to social needs.9 We need more examples like this and have recently published an editorial that provides a range of ideas for how to address equity issues.2

Healthcare is simultaneously complicated and complex. Complicated problems in healthcare often result from the growth in medical knowledge, medications and technology, and poorly designed systems, including systems with accumulated ‘safety clutter’ from prior unsuccessful attempts to improve care.10 These complicated problems can be improved using tools such as Lean, or approaches from human factors engineering. Complex problems arise from systems that are dynamic and often unpredictable; they require different approaches to improvement. For example, Safety-II is an example of a new perspective that argues we need to focus on the complexity of everyday ‘work as done’. The challenge now is to demonstrate how it can be implemented alongside existing safety management practices11 to be effective in improving safety.

We are also interested in understanding the benefits and risks of emerging opportunities to address intractable problems. We welcome submissions using methods, conceptual frameworks, theories or technologies from novel fields or safety-critical industries outside healthcare, provided that these show the relevance for improving the quality and safety of patient care. Some of these may be radical and disruptive in nature—for example, the use of artificial intelligence chatbots in providing medicines information, or remote consultations. These may be especially promising for resource-poor settings but may also pose safety risks. Other opportunities may be innovations for existing ideas, such as adaptations of the ‘Matching Michigan' programme to minimise central venous catheter-blood stream infections.12 We have a strong interest in initiatives that involve patients in problem-solving, through, for example, co-design of safety tools, or shared decision-making.

Implementing solutions in different contexts

It has long been recognised that novel solutions developed and tested in well-resourced academic centres struggle when they reach the rest of a healthcare system. We welcome implementation science submissions that study the underlying mechanisms for why ’best practice’—either new healthcare technologies, or new quality improvement solutions—is not effective in a real-world context. These studies can take a variety of approaches. They might analyse the reasons for why high-quality evidence about new effective technologies does not affect frontline practice or is diffused only weakly. The reasons for non-implementation of evidence from surgical trials is a good example.3 Implementation science submissions might also analyse ‘fidelity’ to improve our understanding or measurement of how far real-world implementation deviates from the ideal.13 A third type of study focuses on getting solutions into everyday work, through tailoring of implementation strategies to a local context. A good example here is clinician performance feedback interventions. We know that a ‘one size fits all’ approach is unlikely to succeed in most contexts, and we recently published a viewpoint that points to ways in which implementation science approaches such as tailoring can overcome this flaw.14 Ideally, these approaches will have generalisable lessons for our broad international audience even as they concentrate on local implementation. For example, a recent Brazilian study of the patterns, appropriateness and outcomes associated with peripherally inserted central catheters demonstrated that assessment criteria and data collection methods previously used only in high-income countries could also be applied in a lower income setting.15 This finding is useful to readers beyond the Brazilian context.

Finally, we know that even when innovations reach the frontline, are tailored to context and implemented with fidelity, their impact can fade over time, or have limited reach across a healthcare system. We welcome submissions on the impact of scaling new approaches to quality improvement12 and how to make them sustainable over time.16

International consensus on basic concepts and measurement

In any field it is difficult to make progress without consensus on basic definitions and measurement tools. There is still disagreement among the various stakeholders in our community about terminology and definitions for fundamental concepts like harm, error, disclosure and resolution.17 Similarly, in 2020, we called for better measurement approaches to adverse events, noting that “we will show progress in patient safety by tracking common, well-defined patient safety problems, not some general measure of all possible harms from medical care, the nature of which will inevitably change over time”.18 Yet we still see incoherence in, for example, the reporting standards and measurement systems used to detect paediatric adverse events.19

Our field is at the point where broad consensus is crucial: how can we consolidate evidence on ‘what works’ if we cannot agree on definitions and measures for the outcomes of interest? This consensus should be globally agreed on and not dominated by movements from one or two countries. Ideally, we would like to receive submissions from credible international initiatives about consensus on definitions and measurement systems.

Continuous improvement of methodology and scientific rigour

We encourage submissions that draw from a wide range of methods, professional and research disciplines, and approaches to improving healthcare. For example, we welcome rigorous and novel work from implementation scientists and quality improvement professionals, from human factors researchers, experts in systems engineering, from health services researchers, the social sciences and health sciences. We try to pick rigorous approaches from different disciplines, to provide high-quality evidence on what works best to improve the safety and quality of care.

As we move forward, we also aim for continuous improvement in the scientific rigour of submissions, and to lift the standard of research carried out in our community more generally. One vehicle we will use for this is our ‘Grand Rounds in Methodology’ series, where “we hope to make healthcare professionals and researchers more aware of the different choices and trade-offs in methods used as well as their impact on generalisability, in order to advance rigour in quality and safety research to benefit patient care and stimulate debate”.20 This series covers three different areas: first, relatively new study designs and research methods such as stepped wedge and adaptive clinical trials; second, methods that are in common use but where mistakes are frequently encountered such as statistical process control charts; and third, broader debates in methodology that are relevant to our field, such as competing risks and sharing of statistical code. So far, we have published three papers in the Grand Rounds series, one on statistical process control, another on realist reviews and, recently, one on implementing machine learning solutions as part of quality improvement interventions; many more will appear in the coming months and years.