Outcome Measures in Sepsis, Revisited*
The ideal short-term outcome measure in pediatric septic shock is patient-centered, easily measured, and clinically important. Post-ICU, long-term metrics such as functional status are clearly valuable outcomes for studies (4), but most investigators will not be able to practically obtain them, and finding a short-term surrogate that relates to long-term morbidity remains an important struggle for our field.
The Vasoactive-Inotropic Score (VIS), first proposed in 2010 for use in children postcardiac surgery (5), has since been extensively used to quantify the cardiovascular support provided in critically ill children with a variety of syndromes (6, 7). As a tool that is easily calculated at the bedside incorporating a variety of vasoactive and inotropic agents, it is a continuous variable that accurately represents the intensity of hemodynamic support. For the cardiac surgical population, this was a significant leap forward in quantifying the hemodynamic needs of patients at fixed timepoints after their surgeries, allowing for intelligent comparisons to be made between patients and institutions. Its use as a predictor of outcome has been shown, where maximum scores in the first 24 hours after surgery are associated with composite outcomes of morbidity and mortality (8).
Sepsis, however, is different. Children with sepsis are much more heterogeneous, with great variations in timing of presentation, nature of disease, presence of comorbidities, and management strategies. Standardizing an outcome measure based on discrete timepoints and intensity of provider therapy will be difficult, much more so than in the relatively homogeneous cardiac surgical population.
In this issue of Pediatric Critical Care Medicine, McIntosh et al (9) took on this challenge. They used their quality improvement database for children with sepsis in their low-mortality center, and attempted two feats: 1) to determine whether a relatively early VIS correlated with their primary outcomes of ventilator days and length-of-stay and 2) to determine the best timepoint to measure VIS for that first purpose. A previous article, in a high-mortality region, documented an association between a high VIS measured at points within the first 48 hours of admission and ICU mortality (10). McIntosh et al (9), with a much lower mortality in their cohort, used the alternate primary outcomes of ICU length-of-stay and ventilator days, and a composite secondary outcome of cardiac arrest/extracorporeal membrane oxygenation (ECMO)/in-hospital mortality in 138 consecutive patients over 3 years.
The VIS, measured as a continuous variable at 48-hour post-ICU admission, seemed most associated with their primary outcomes, compared with measurement at earlier junctures. This association was independent of the measured Pediatric Index of Mortality-3 score, suggesting some specificity of 48-hour determination of cardiovascular support requirements. Measurements of VIS at 12 hours were associated with their secondary composite outcome of ECMO/cardiac arrest/death. Wisely, they did not categorize the VIS into arbitrary thresholds of “high” and “low,” as in prior work, ensuring maximal power for their calculations (11).
Of interest, contrary to the postcardiac surgical population, the maximum VIS achieved during ICU admission was only very weakly correlated with their primary outcome.