© 2005 American Thoracic Society doi: 10.1164/rccm.2411004
Dissecting Sepsis Outcomes in Critically Ill ChildrenGeorge Washington University School of Medicine Washington, DC The ability of physicians to predict the outcome of intensive care unit (ICU) patients is highly variable. The science of severity of illness measurement has revolutionized critical care practice by providing a more objective means of assessing illness severity and for prognosticating outcome. There are now global models available for all age groups and a number of measures for specific disease entities (1–5). The ability to quantify objectively severity of illness in the ICU is important because it allows ICUs to compare themselves with respect to clinical and performance outcomes (5). In the setting of clinical trials, severity measures permit the stratification of risk for different treatment arms. For populations of patients with a particular condition, severity of illness models provide the opportunity to understand changes in prognosis as new treatment strategies become available (5). However, severity of illness models also have a number of limitations that require them to be used cautiously in the patient care arena (5, 6). Sepsis is an important clinical problem in critical care because it occurs commonly and has a high mortality rate (7). Multiple organ dysfunction syndrome (MODS) is associated with sepsis and also has poor outcomes. Although it makes sense conceptually that sepsis and MODS occurring together would be associated with increased mortality, at least one report in the literature failed to demonstrate such a relationship (8). This may in part be related to the traditional approach of cataloguing organ dysfunction by using a nominal variable that indicates the presence or absence of organ failure. Providing an objective measure that more completely analyzes septic states and organ dysfunction may be valuable to clinicians caring for these critically ill patients. Leclerc and colleagues (9) in this issue of AJRCCM (pp. 348–353) advance prior work in pediatric MODS by considering the influence of both sepsis and progressive organ dysfunction on mortality. The authors prospectively investigated the cumulative influence of organ dysfunction and the septic state on mortality, using multivariate survival analysis of nearly 600 children admitted to three pediatric ICUs in 1997 with both sepsis and organ dysfunction. A scaled MODS score was used to provide estimates of the hazards of mortality for each organ dysfunction and diagnostic category. This score, known as the Pediatric Logistic Organ Dysfunction (PELOD) scoring system, has been validated in a large, multicenter, observational cohort (10). The authors' research was able to demonstrate several important findings relevant to both pediatric sepsis and MODS. First, the authors validated that septic states are common in the pediatric ICU environment and occur early in the ICU course, with a range of severity from systemic inflammatory response syndrome (SIRS) to septic shock. Second, as the severity of the septic state worsens, mortality also worsens. Third, sepsis and organ dysfunctions are synergistic in worsening outcomes. In fact, the authors were able to ascribe a hazard ratio for each organ system dysfunction according to the severity of the septic state. These findings as a descriptive effort are meritorious. However, as with any research, the extent of limitations needs to be considered to accurately appraise the work. First, the study is limited to three ICUs. Of importance, the three units are heterogeneous with respect to patient characteristics, including the number of admissions (range, 106–336), median patient age (5–44 months), surgical patient volume (13–58%), and median Pediatric Risk of Mortality (PRISM) scores (2–8). The units are so heterogeneous, in fact, that combining these units into one report without the opportunity to control for unit-specific effects creates potential problems with the clustering of results by unit. This is clearly demonstrated in Table 1 of the article, in which the three units are directly compared. Second, the population is limited to about 600 patients and 50 deaths. The population is interesting in that illness severity as estimated by PRISM scores is relatively low for an ICU population. Furthermore, the sample sizes for the most severe septic state (shock) are small, which prohibits the ability to make generalizable statements regarding the most severely ill patients. The hazard ratios with each sequential organ dysfunction are impressively large, but the validity of these numbers is uncertain with so few patients representing the most severe end of the illness spectrum. The value of these hazard ratios is derived from their basis on easily quantifiable disease-specific outcomes that have face validity. However, their ability to perform outside of the setting in which they were constructed reduces the value of this report to other ICUs. Third, the authors grouped the SIRS and sepsis groups together because of similarities in outcome. SIRS and sepsis represent two of the largest groups in the study, but are heterogeneous populations. By combining these groups, the authors may have missed subtle outcome differences related to organ function that could be used to benefit patients. As demonstrated in Table 2 of the article, the relationships between progressive organ dysfunction and mortality in these two groups vary. Patients with SIRS and sepsis with fewer than one organ dysfunction each have an observed mortality of 0 (9). With three-organ dysfunction, the septic patients had a higher mortality than those with SIRS, but this relationship changes for patients with four and five organ dysfunctions, among whom higher mortality occurs with SIRS. One of the major benefits of using a score like PELOD is that intermediate outcomes, such as organ failure, can be identified. By combining two of the major septic states, the authors have diluted the value of their methodology. Although these limitations are notable, it should be recognized that the authors have contributed to our understanding of the effects of MODS and sepsis on mortality. This relationship is clinically feasible and the use of a scaled score for organ dysfunction provides additional insight that advances prior efforts in this area. This scaled score may provide value in assessing outcomes from different treatment arms as we attempt to improve the treatment of sepsis with the application of new therapeutic strategies. In addition, there are now opportunities to better understand the relationship between organ failure and sepsis as well as their combined effects on mortality. This is important because sepsis and organ dysfunction are common occurrences in children hospitalized in the pediatric ICU. However, until some of the concerns regarding bias and small sample sizes are addressed, the ability to use these scales prognostically in individual patients should be avoided (11, 12). FOOTNOTES Supported in part by Agency for Healthcare Research and Quality grant KO-8 HS-14009 (A.D.S.). Conflict of Interest Statement: D.C.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; A.D.S. has received a grant fund in patient safety from the Agency for Healthcare Research and Quality and has received funding from MedImmune and the Child HealthCare Corporation of America. REFERENCES
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
