Have Outcomes from Sepsis Improved?

Recent studies have suggested that outcomes for patients with sepsis have improved over the years. As early as 1998, in a review of studies examining patients with septic shock published between 1958 and 1997, Friedman et al. reported a decrease in hospital mortality rates from about 65% to about 42%. In another early study, Martin et al. reported that in-hospital mortality rates for patients with sepsis admitted to a sample of U.S. hospitals decreased from 28% for the period 1979 to 1984 and 18% for the period 1995 to 2000. More recently, Stevenson et al. used data from the control arms of randomized clinical trials in patients with sepsis published between 1991 and 2009 and reported a 3% annual decrease in 28-day mortality rates ( P = .009). The same authors and others have reported similar trends for in-hospital mortality when using administrative hospitalization data in the United States and other countries. Using data from the Australian and New Zealand Intensive Care Society adult ICU patient database, Kaukonen et al. reported an absolute decrease in the hospital mortality rate of sepsis from 35% in 2000 to 18.4% in 2012; after logistic regression analysis, the odds ratio (OR) for mortality was 0.49 (95% confidence interval [CI], 0.46 to 0.52) in 2012 with 2000 as reference.

Taken together, there is, therefore, some evidence of improved outcomes from sepsis over the last couple of decades ( Table 44-1 ). Nevertheless, the apparent extent of the decrease in mortality should be interpreted with some caution. Indeed, increased awareness of sepsis, changes in the code definitions used to classify the disorder, and altered reimbursement strategies have likely led to an inclusion of an increased number of patients with less severe disease and, hence, inherently lower risk of death, in studies on sepsis; this effect certainly accounts for some of the reported temporal increase in the number of septic patients—including less severe cases—with concurrent decrease in mortality.

What Has Not Worked?

Over the years, our understanding of the pathophysiology of sepsis has improved so that many of the complex responses to infection and how they interact to cause sepsis are now well detailed and defined. Multiple pathways and molecules have been identified as potential targets for therapeutic intervention; however, despite more than 100 randomized controlled clinical trials of sepsis-modulating therapies, no effective intervention has been identified. Clearly then, this approach to improve survival has not worked. There have been many putative explanations for these apparent “failed” trials, including discrepancies arising when preclinical models and experimental data are translated to the clinical arena; issues with the in vivo efficacy of the intervention under examination; concerns about the dose and timing of the intervention; and problems with clinical trial design, including choice of outcome measures. Perhaps the key problem, though, has been in the selection of patients for these studies. Lack of a clear and specific definition or marker of sepsis has led to the inclusion of very heterogeneous groups of patients. Patients with different degrees of disease severity, different sepsis sources and causative microorganisms, different genetic backgrounds, and different comorbidities and ages have all received the same intervention. Many studies also included multiple centers with an associated variability in standards of care, resource availability, and staff training. Moreover, it has become apparent that patients have different types of immune response—both proinflammatory and anti-inflammatory responses are present simultaneously—and the balance between these two forms may determine a patient’s response to treatment. This has rarely been taken into consideration when clinical trials are designed. In a trial that includes such heterogeneous groups of patients, a single intervention may be of benefit in some but harmful in others so that the overall study outcome may not accurately reflect the true efficacy of the therapeutic agent had it been tested in a more select population. For example, a patient with a primarily proinflammatory response is unlikely to respond to an agent that further promotes inflammation; thus administration of granulocyte colony-stimulating factor (G-CSF) to all patients with septic shock was not associated with improved outcomes. Similarly, giving an anti-inflammatory agent to a patient who is already immunosuppressed will probably not be of benefit. Indeed, in many of the studies of immunomodulatory agents that showed no overall improvement in outcome, beneficial effects were identified in certain subgroups.

Other specific aspects of patient management have also not consistently been shown to be effective. An early goal-directed therapy protocol reduced mortality in a selected group of patients at a single center but had no beneficial effects on outcomes in two larger, multicenter studies. Similarly, tight blood glucose control improved outcomes in a single center study on critically ill surgical patients but not in a more general population of ICU patients. Glucocorticoid therapy reduced the risk of death in one study in patients with septic shock, but these effects were not confirmed in later studies.

Single interventions in heterogeneous groups of “septic” patients have therefore clearly not worked. Improving patient characterization so that those patients who are most likely to respond to the intervention(s) in question can be identified and studied is necessary for future clinical trials in sepsis therapeutics.

What Has Worked?

Despite the lack of specific sepsis treatments and some problems with diluted data, patient outcomes from sepsis have improved over the years. Therefore if single specific interventions have not been effective, what has worked? It is logical to invoke two major factors in these improved outcomes: (1) the enhanced awareness of sepsis as a possible diagnosis and realization of the importance of early recognition and management and (2) a gradual improvement in the general process of care for these, and indeed all, critically ill patients. Taking the former aspect first, early effective antibiotic treatment, infectious source removal, adequate fluid administration, and vasopressor and organ support have all been associated with improved outcomes. Guidelines with recommendations for best patient care, stressing the need for rapid institution of these practices, have been written by teams of experts, and bundles of care items (including measurement of blood lactate level, early administration of broad-spectrum antibiotics, administration of fluids when hypotension is present, and administration of vasopressors for hypotension that does not readily respond to initial fluid resuscitation) have been developed. Compliance with these bundles has been associated with improved outcomes in different ICU settings, although intensivists should not be restricted by specified time limits and all aspects of these bundles should be performed as rapidly as possible. The use of multidisciplinary sepsis response teams has been suggested to improve the initial stabilization of patients with sepsis, ensuring that all aspects of management can be performed rapidly. A specially equipped and staffed room or “shock lab” could similarly improve early management in these patients.

In terms of process of care, of the many aspects that have seen gradual change over the years and led, in combination, to improved patient outcomes in all critically ill patients, including those with sepsis, four merit specific discussion. The development of intensive care as a specialty in its own right with trained intensivists familiar with the complexities of critical illness has contributed hugely to the ongoing improved process of care. First, intensivists have generally become less invasive and less aggressive in some aspects of their patient management. They have come to understand that many of the seemingly pathophysiologic effects of sepsis are, in fact, beneficial and should not necessarily be “treated” or “normalized.” The use of interventions that have been associated with poorer outcome has gradually been reduced and even eliminated. Thus, fewer transfusions are given, patients are fed less, tidal volumes have been reduced, and sedation has been minimized. Second, intensivists have come to appreciate the unique circumstances surrounding each patient and have thus individualized treatment rather than manage all ICU patients in the same way. Conversely, intensivists have standardized critical aspects of care by introducing guidelines and protocols so that key elements are less likely to be forgotten or mismanaged. This dichotomy can, in some circumstances, become problematic. Although protocols can improve the delivery of care when quality is suboptimal, especially when there is a shortage of well-trained staff, they may be too rigid in many centers where care is already optimal and may limit intensivists’ ability to account for the importance of individual patient factors; here, checklists may be a better approach. Third, intensivists have realized the importance of multidisciplinary teamwork within the ICU setting, moving from a rather paternal, physician-directed approach to patient management and decision making that is much more inclusive, with input from all members of the ICU team, including nurses, physiotherapists, nutritionists, and pharmacists. Good teamwork can help reduce medical errors and improve job satisfaction, as well as improve patient outcomes. One of the key aspects of good teamwork is good communication, and this concept extends also to patients and their relatives. Patients, whenever possible, and next of kin are now informed more openly of patient progress, treatment options, and likely prognosis. End-of-life decisions in particular are now discussed more candidly and clearly with families, and patients increasingly share in the decision-making process. Fourth, realization of the importance of early recognition and management of critical illness has led many hospitals to extend the ICU beyond its physical four-wall structure by creating medical emergency teams or ICU outreach teams. These consist of trained intensivists, nursing staff, or both who can assess and initiate management of patients on the general ward before they deteriorate to the point where they require ICU admission. Critical illness generally starts some time before ICU admission, and the severity of illness could potentially be limited by early intervention, thus improving patient outcomes. Similarly, early patient mobilization has largely improved the convalescent phase.