Improving Sepsis Care: Title and subTitle BreakThe Road Ahead

In recent years there have been unprecedented advances in the understanding of the epidemiology, pathophysiology, and treatment of sepsis syndrome.^{1 - 3} This work has culminated in several clinical trials demonstrating the efficacy of targeted interventions to improve sepsis-related outcomes.^{4 - 6} These interventions include not only novel therapeutic agents such as drotrecogin alfa but also treatments directed at improving the way more traditional therapy is delivered, such as early resuscitation and low-tidal volume ventilation for acute lung injury.^{4 - 6}

Unfortunately the gaps between evidence and practice have long been huge.⁷ Indeed, most available data suggest that clinical trial and observational study results have not yet changed clinical practice in sepsis care. Few emergency departments have implemented protocols for early resuscitation of patients with severe sepsis, delayed and inappropriate antibiotic administration remains common, and many patients with acute lung injury receive mechanical ventilation with potentially injurious tidal volumes.^{8 - 10}

Numerous obstacles get in the way of implementing clinical evidence. Clinicians may be unaware of published evidence, disagree with practice guidelines, or be unable to effect change due to environmental and structural barriers.¹¹ These challenges are particularly salient in sepsis care, which requires dedicated efforts between multiple disciplines and coordination of care throughout the hospital, all in a setting in which time to treatment is central. Comprehensive strategies are needed to standardize practice, improve care processes, and optimize outcomes for this high-risk patient group.

Recent evidence suggests that grouping care practices together into “bundles” may be an effective method to improve outcomes for complex diseases such as catheter-related bloodstream infections, ventilator-associated pneumonia, and even sepsis.^{12 - 14} But it has proved extremely challenging to take complex care improvement programs and disseminate them broadly across a region, state, country, or across national boundaries.

In this issue of JAMA, Ferrer and colleagues¹⁵ report the findings of an ambitious, nationwide effort to improve the quality of care for patients with severe sepsis and septic shock. A total of 59 intensive care units (ICUs) participated in the program, representing 21% of all ICUs in Spain. Each hospital received an educational intervention based on the Surviving Sepsis Campaign, an international program designed to increase sepsis awareness and develop and disseminate practice guidelines.¹⁶ The intervention consisted of identifying a local clinical champion, assembling a multidisciplinary team with broad stakeholder involvement, a baseline performance audit, educational lectures, and guideline dissemination to physicians and staff in the emergency department, in hospital units, and in the ICU.

In the period following the intervention, patients were more likely to receive early appropriate antibiotic therapy, adequate fluid resuscitation, and documented consideration of drotrecogin alfa and low-dose corticosteroids. Survival also improved, with statistically significant reductions in both hospital and 28-day mortality. Improvement in survival was greatest in hospitals with the poorest baseline performance. These performance gains provide an important process-outcome link in support of the sepsis guidelines because some of the elements of this campaign have not yet been strongly linked to outcome in patients with severe sepsis.

The intervention was associated with important process and outcome improvements even though it was relatively simple. Didactic teaching and passive guideline dissemination are not the most effective methods of behavior change.¹⁷ The investigators did not include some of the more effective methods for implementing evidence-based practice, including academic detailing, computerized reminders, and repeated audit and feedback.^{18 - 19} Additionally, the intervention was homogeneous across sites, with no attempt to customize the program based on local cultures or specific organizational barriers.²⁰ The fact that performance improved even after this type of intervention is probably due to poor compliance and high mortality at baseline.

The study has limitations. Most notably, the absence of a cluster randomized design or even concurrent controls leaves open the possibility that temporal trends or changes in case-mix led to the observed association. Nonetheless, the sheer scope and scale of this initiative should not be discounted. The campaign in Spain represents the culmination of a multi-year effort involving coordination between the investigators, the Spanish Society of Intensive Care Medicine and Coronary Care Units, and the participating sites. Simply the fact that a national professional society undertook a project to improve its country's health and was able to achieve improvements in both process and outcome makes this work unlike any previous quality-improvement initiative in hospital medicine. Perhaps the greatest lesson of this study is that through multicenter collaboration it is possible to meaningfully effect change in the quality of hospital care, not just locally but across an entire nation.

This work also highlights several of the challenges in hospital-based quality improvement. First, only a minority of Spanish hospitals were willing to participate in the program, and several dropped out. Hospitals were not asked to justify their nonparticipation, but many likely either could not identify a local champion or were unwilling to devote scarce resources for a collaborative quality-improvement effort. Innovative methods of quality improvement that do not require local champions and are more easily exportable to multiple sites, perhaps involving use of information technology, may be more effective.

Second, baseline performance was relatively poor despite the presence of 24-hour intensivist staffing at all sites. Intensivist physician staffing is consistently linked to improvements in the process and outcome of critical care.^{21 - 22} Yet it is clear that intensivists alone are not a cure for poor quality. Aggressive quality-improvement efforts are needed even in optimally staffed ICUs.

Third, the increases in guideline adherence were modest, with some process measures increasing by only a few percentage points. Even after the intervention, performance on many measures was far below what would be considered ideal. Nor were all improvements sustained in a subset of hospitals that measured process and outcome for a year after the intervention; many measures returned to baseline rates. The benefits of this type of intervention might well increase over time if organizations develop effective ways to ensure that the bundles are implemented, but might also wane if the focus shifts to other areas. Future research should address how to better maintain process improvements, including the role of incentives and ongoing performance measurement.

Ferrer and colleagues supply powerful evidence that broad-based quality improvement in sepsis care is feasible on a national scale. The data also suggest that delivering a bundle of care effectively for patients with sepsis may be as or even more important than developing new therapies. Indeed, the absolute risk reduction in hospital mortality observed in this study would translate to an impressive number of lives saved if this type of intervention were successfully implemented on an international scale. The science of quality improvement must include not only development of effective measures, but also evaluation of what techniques for spreading and maintaining them are most effective.

Furthermore, this study should be a wake-up call to policy makers, a challenge to the leaders of professional societies, and a road map for the path ahead. No longer is it acceptable to simply publish practice guidelines and hope that quality improvement happens at the local level. Development of these guidelines should be followed by rigorous testing, and, when results are positive, by dedicated regional, national, and even international implementation efforts. Such broad-based efforts are needed to achieve population-level benefits from interventions known to be effective.