This study describes the structure and implementation of a nationwide surgical trainee-led QI programme to increase the use of perioperative risk scoring in emergency laparotomy using the active implementation framework. Over the duration of the programme, the percentage of patients undergoing emergency laparotomy who received perioperative risk scoring increased from 0-11% during the exploratory phase to 35-100% during the full implementation phase. This change was consistently maintained throughout the full implementation phase of the programme.
Two large scale, prospective Emergency Laparotomy QI programmes from the UK have recently been published, from the Emergency Laparotomy Collaborative (ELC) group(30) and the Enhanced Peri-Operative Care for High-Risk Patients (EPOCH) trial group(31). The ELC study involved the implementation of a 6-point care bundle in 28 National Health Service (NHS) hospitals and found a reduction in the unadjusted mortality rate and the length of stay as well as changes in some of the care bundle metrics. This programme used the Institute for Healthcare Improvement (IHI) Breakthrough Series collaborative as an implementation framework. Although, perioperative risk assessment was not part of the care bundle, it demonstrated that QI methodology can improve outcomes in emergency laparotomy at scale(30). The EPOCH trial intervention consisted of an extensive 37-point evidence-based care pathway in 15,873 patients in 93 NHS hospitals. Overall, the study investigators found that there were modest improvements in 10 patient-level process measures, however this did not translate to overall outcome measures including overall survival, 90- and 180-day mortality, length of hospital stay or frequency of readmission(31). There were wide variations in intervention fidelity between hospitals, with differences in the processes that individual teams tried to change, the rate of change and the eventual success. In a hospital-level evaluation of the EPOCH trial, no hospital reliably implemented all 10 processes. Out of 93 participating hospitals, 80 provided sufficient data for analysis generating 800 process measure charts from 20,305 patient admissions over 27 months. Overall, only 279 of the 800 processes were improved(32). These findings show that change at scale and the context in which change is being implemented is far more complex than initially anticipated. Context is central to the success (or failure) of QI programmes.
Successful implementation is not only influenced by context but also by the content of the intervention and the process of implementation. A number of reports have been published on the effects of small-scale QI projects to improve outcomes in emergency laparotomy. The ELPQuicC group assessed the implementation of a bundle of 5 interventions in 4 hospitals and reported a reduction in risk-adjusted mortality (risk ratio 0.61, 95% confidence interval 0.45-0.84)(20). The difference in findings compared to the EPOCH trial may relate to the simpler intervention, but also a stronger pre-existing relationships between staff leading implementation in these early adopter hospitals. The EPOCH trial set ambitious targets in hospitals in which there may have been a less favourable organisational readiness for change that in the ELPQuicC hospitals. Reports of QI programmes in other clinical areas have delivered mixed results(33-35), suggesting that more focused, discrete clinical interventions might be more successfully implemented than interventions that include a larger number of clinical processes.
A mixed-methods process evaluation of the planning, delivery and reception of the EPOCH intervention at hospital level found that fidelity of the intervention showed considerable variability across individual hospitals and only 11 of 37 care pathway processes that more than half of respondents reported attempting to improve(36). Analysis of the qualitative data for the trial suggests QI leads were often attempting to deliver the intervention in challenging contexts and identified barriers including the social aspects of change such as engaging colleagues, limited time, limited organizational resources and challenges in collecting data(36).
The model of trainee-led QI programmes are relatively sparse in the literature. There have been successful trainee-led programmes aimed at improving patient satisfaction(37), increasing compliance with venous thromboembolism prophylaxis(38), reducing opioid analgesia discharge prescriptions(39) and other programmes(40, 41). This study demonstrates that within a structured programme, surgical trainees can deliver improvement at a national level. Postgraduate surgical trainees in resource-limited healthcare systems are adaptable and resourceful and with the correct support an ideal group of healthcare professionals to drive change at local level. In addition, trainees are front line staff that have the potential to influence the intervention, engage staff and counteract resistance and suspicion about externally led change(42). In the context of limited organization resources and potentially challenging data collection, a simple, explicitly designed measure with data collection, monitoring and a feedback system incorporated into improvement activities from the outset. The primary outcome of a documented preoperative risk scoring was chosen as a balance between a credible measure by participants without being unduly burdensome, or difficult to interpret.
Adaptability and trialability is often essential in ensuring QI interventions can fit within different contexts and is a driver of success in our programme. However, fidelity to key parts of an intervention is also important to maximise the likelihood of success(43). Although fidelity was not directly assessed in our programme, it is likely there was considerable variability in the local implementation of the intervention and this may impact long term sustainability of improvements achieved.
Central to the success of the programme are the social aspects of improvement which are as likely to be as important as more technical aspects, such as data analysis and feedback. Similarly, a central expert implementation team oversaw the development of local programme teams building local QI capacity and serving as an accountable structure to move through the stages of implementation successfully. Local programme leads capitalised on social QI strategies to drive change through utilization of a participant messaging platform as well as the programme leadership through use of monthly participant teleconferences. Building and maintaining effective social relationships is time-consuming and challenging, particularly senior colleagues and hospital administration(44). This may be even more challenging for trainees who usually have temporary training appointments. Preoperative risk assessment is not a magic bullet. Developing consensus around the value of a formalised process of preoperative assessment required some to engage senior surgeons of the value of the selected intervention(45).
Given that this was an unfunded project, the resources and time available to coach local QI leads was limited in comparison to other reported QI interventions such as the IHI Breakthrough Series Collaborative model(46). Our programme was designed as a parsimonious intervention, with limited face-to-face meetings, so that it could be adapted and replicated widely if found to be successful. A higher intensity coaching programme might have led to greater intervention fidelity, however, it is uncertain if this is always the case(47, 48).
There are some inherent limitations to our study which need to be acknowledged. The primary outcome was a documented preoperative risk score for patients undergoing an emergency laparotomy at each site, however, we did not evaluate changes in patient outcomes associated with the intervention. Variations in the denominator for the bi-weekly measures sometimes interfered with signals in the data. For example, in a 2-week period with a small denominator, fewer patients scored may create a data point that breaks a signal that would otherwise indicate a move towards improvement. In combination with the time-bound nature of the analyses, this may have led to some real-world improvements not being identified using the run charts.