Making and testing change in healthcare

Greg Ogrinc and colleagues call for greater exploitation of the synergies between quality improvement and implementation science in improving care

 This is an edited version of an article that originally appeared on The BMJ

Improving the quality of healthcare is complex. It requires input not just from healthcare providers but also from patients and families to identify gaps, develop meaningful interventions and ensure that interventions improve care and outcomes, considering value from their perspective. Closing gaps in healthcare quality, improving workflows and implementing evidence-based interventions requires change, but not all changes are successful, and most come with unintended consequences.

Numerous approaches are available to making changes in healthcare systems such as lean, six sigma, the model for improvement, healthcare delivery science and implementation science; these are usually used in isolation, although there is some overlap in their approaches, particularly quality improvement and implementation science. When looking to build and disseminate knowledge about making change, collaboration between approaches might help create changes more successfully and efficiently.

Interdisciplinary tensions

Over the past several years we have recognised a tension and, at times, a competition between quality improvement (QI) and implementation science (IS), two commonly used systematic approaches to improve the quality, safety and value of healthcare services, and to disseminate what is learnt from those efforts. This tension is unnecessary and wasteful when so many gaps in healthcare quality need to be addressed. QI focuses on the highly relevant work within a particular context, while IS focuses on framing the work to make findings generalisable. Both approaches are important, and there is considerable overlap from which each can learn. Failure to recognise the overlap can lead to replication of interventions, delay in the dissemination of effective interventions and missed opportunities to work together to improve healthcare. Others have also recently noticed overlap in QI and IS such as the potential to use both to improve cancer care.

The response of healthcare systems to the COVID-19 pandemic exemplifies this challenge. Early in the pandemic local systems experienced a rapid influx of patients with COVID-19 and dwindling supplies of personal protective equipment (PPE). Institutions relied on sound QI methods to determine how to solve the particular problem of PPE in their particular contexts. While this was necessary and helpful, there was, perhaps, a missed opportunity; if IS methods had been used to help solve those problems, it might have been easier to share answers with others.

Both QI and IS bring a rich base of knowledge and skills. Both are needed but their potential summative effects have been underused during the COVID-19 pandemic, partly because these fields see themselves as competitors and not collaborators.

QI and IS approach change from different philosophical underpinnings, yet we feel they share similarities that suggest combining their lenses would be beneficial. While QI comes from system operations, and IS from behavioural sciences, both recognise that changes occur in a specific context and are affected by the context itself, requiring that each context be considered unique. The outcomes of interest in QI are generally improving quality of care – safety, timeliness, effectiveness, equity, efficiency and patient centredness. The outcomes of interest in IS generally include the uptake and application of evidence-based care with attention to acceptability, cost and feasibility. Both fields focus on disseminating findings to others through peer-reviewed publications. Overall, this tension has been accurately described as the work of moving evidence into practice.

Different approaches to change

Making and documenting changes to interventions is difficult, and reporting of QI and IS varies, limiting impact. The Standards for Quality Improvement Reporting Excellence (SQUIRE) were developed to improve the reliability of reporting among those using QI, IS or any of the other approaches to improvement.

The challenge of reporting is prominently manifest in how the intervention changes in response to the local context. An intervention has an initial structure, but the intervention is, typically, modified throughout the process of improvement or implementation to make it more effective in a particular context. Both fields use imperfect approaches to capturing the data related to these processes, and each has something to offer the other.

In QI changes are widely promoted to be accomplished through ‘tests of change’ to predict, test and assess the effect in the local microsystem. Careful use of tests of change, such as through plan-do-study-act (PDSA) cycles, are viewed as key to learning about the microsystem and the context to inform the change process. The strength of this approach is the importance placed on the microsystem and context and, in this way, it may inform the IS field, which does not strongly emphasise the use and reporting of recursive change.

In IS the concept for modifying the intervention is referred to as ‘adaptation’ and is recognised as key to identifying how to spread effective interventions to new contexts, rather than on making the change work in one specific context. Making, assessing and reporting these adaptations are viewed as an essential part of generating knowledge that can be readily shared with others. The strength of IS is in the methods and approaches to the assessment and reporting of adaptations, and these could be usefully applied to QI, which does not emphasise spread to other contexts as strongly.

In IS models such as the consolidated framework for implementation research (CFIR), provide a framework of domains of context; these help inform the design of interventions and are especially useful for formative evaluations of change. CFIR domains, for example, are prespecified contextual factors and sub-constructs that include the adaptability and trialability of the intervention. CFIR also recognises that PDSAs are one way to adapt an intervention to a specific context. Importantly, both QI and IS note that there is bidirectional interaction, with the intervention affecting the context and the context affecting the intervention.