Reality check for checklists.

Catheter-related blood stream infections in the intensive care unit (ICU) are common, costly, and potentially lethal. The Dec 28, 2006, issue of The New England Journal of Medicine reported that an evidence-based intervention in 103 intensive care units in the Michigan Keystone ICU programme had resulted in a large sustained reduction in rates of these infections. The study was widely reported in the popular media and elsewhere as a triumph of the “simple checklist” as a solution to patients’ safety problems. Yet the widespread interest in this study is a dual-edged sword. It was a great story. Science often needs to be simplifi ed for the lay public. The problem is that the story may well have been oversimplifi ed. The emphasis on checklists is a Hitchcockian “McGuff an”, a distraction from the plot that diverts attention from how safer care is really achieved. Safer care is achieved when all three—not just one—of the following are realised: summarise and simplify what to do; measure and provide feedback on outcomes; and improve culture by building expectations of performance standards into work processes. We propose that widespread deployment of checklists without an appreciation of how or why they work is a potential threat to patients’ safety and to high-quality care. Attributing the reduction of infection in the Keystone programme solely to the use of checklists is an easily made but crucial mistake. Checklists are a good way of making certain that tasks get done, as anyone who has taken a shopping list to the supermarket can testify. If wise, checklists can help workers perform a task by reducing ambiguity about what to do. Of course, determining the best way of proceeding in a complex health-care setting is not as straightforward as producing a prompt to remember the milk. But fi guring out what should form the content of a checklist for a clinical problem is a nonetheless achievable ambition: there are well-defi ned processes for identifying and synthesising research evidence. For the Keystone programme, interventions with a potential to improve outcomes were identifi ed, and the fi ve procedures that had the strongest evidence and the lowest barriers to implementation were selected and converted into a standardised checklist. But checklists, even if based on rigorous evidence, have never penetrated medicine in the way they perhaps ought to have. The reasons for this are primarily social and cultural. In part, the way that physicians are socialised creates resistances and interferences to the use of checklists. Some come to feel that checklists undermine their claims to expertise, are infantilising, and an unnecessary impediment to the swift decision making and action required for eff ective care. How to understand and disrupt these deeply entrenched norms is a much greater challenge than identifying the components of a checklist. The mistake of the “simple checklist” story is in the assumption that a technical solution (checklists) can solve an adaptive (sociocultural) problem. To improve safety, health care needs to get the technical and adaptive work right. Without attention to adaptive work, checklists would probably suff er the same fate as guidelines—often left unused, even when very robust. Summarising evidence is a necessary but not suffi cient step for translating evidence into practice. Evidence summaries need to be combined with an understanding of, and a strategy for, mitigating the technical and social/political and psychological (even emotional) barriers to using the evidence, and with feedback about performance. Emphasising checklists as the explanatory mechanism for the reduction in catheterrelated infections obscures the complex labour necessary to create a collective local faith in checklists. How support was mobilised for coordinating work around infection control is the real story of the Keystone ICU project. What happened in Michigan involved the creation of social networks with a shared sense of mission, whose members were each able to reinforce the eff orts of the other to cooperate with the interventions. Implementing the entire programme occurred over 9 months—it was not simply the case that the units were handed the checklist and immediately fell in line. The work was arduous and often laden with emotions. Before ICU units were allowed to take part in the intervention, each hospital had to assign a senior executive to work with participating units. Each ICU was required to identify a physician and nurse team leader. The executives were required to meet monthly with unit Co bi s