Let Us Not Forget Early Mortality in Pediatric Sepsis

Death from sepsis, while tragic and far too common, has been surprisingly enigmatic. Although many patients who succumb to sepsis have refractory shock and most exhibit multiple organ dysfunction syndrome, the mechanisms underlying treatment failure and death remain largely elusive (1, 2). Recently, several studies described the timing and mode of death in children with sepsis that have improved our understanding of sepsis-related mortality. In particular, three studies conducted around the same time in the United States (US), United Kingdom (UK), and Australia/New Zealand each reported a similar bimodal distribution in the timing of death among children with sepsis, with 25–55% of deaths occurring “early” (within 1 day of pediatric intensive care unit [PICU] admission or sepsis recognition) and 29–51% after 7 days (3–5). In all three studies, children who died early were more likely to be previously healthy, have refractory shock, and succumb to unsuccessful cardiopulmonary resuscitation (CPR) rather than withdrawal of life-sustaining therapies compared to those who died later in their course of illness.

In this issue of Pediatric Critical Care Medicine, Dr. Johnson and colleagues add to these observations with the largest study to date of deaths among critically ill children treated in a PICU for severe infections (6). They included 130 US and Canadian hospitals in the Virtual Pediatrics Systems (VPS, LLC) database between January 2009 and December 2014, which yielded data from 106,461 admissions with at least one infection-related diagnosis code. Overall, 4,240 patients died (4.0% mortality rate). Not surprisingly, septic shock, severe sepsis, and sepsis were among the most common diagnosis codes assigned to non-survivors, reflecting sepsis as the usual pathway to death among infected children. Six hundred thirty-five deaths (15.0%) occurred “early” (within 1 day of PICU admission) and 50% occurred after 7 days. Of the subset of 1,553 patients who died with severe sepsis or septic shock, 328 (21.1%) died within 1 day of PICU admission. As in the studies by Cvetkovic et al, Schlapbach et al, and Weiss et al (3–5), children who died early were more likely to have septic shock (rather than severe sepsis or another infection-related diagnostic code without shock) and require CPR and less likely to die following withdrawal of life-sustaining therapies compared to those who died later.

Although this report by Johnson et al. lacked the data granularity accessible in prior studies, a major strength was its large sample size spread over a substantial number of hospitals diverse in geography, size, and patient composition. However, because VPS is a PICU-based data source, patients who died very early—prior to PICU admission—would be excluded from this study, whereby prior studies included such deaths (3, 5). It is, therefore, quite notable that their observation of one out of every five deaths among critically ill children with sepsis occurring within 1 day of PICU admission confirmed the findings from studies with smaller sample sizes. Moreover, key characteristics of early deaths—i.e., high rate of shock and frequent, unsuccessful CPR—were also corroborated in this larger study (although data on comorbid conditions were not reported).

The consistency of these findings across countries with different healthcare systems emphasize what appears to be an epidemiologic truth—many children still die from sepsis very soon after hospital admission. The reason for such early deaths is likely multifactorial, stemming from a combination of rapid progression to severe illness, refractoriness to usual resuscitative therapies, late presentation, and diagnostic delays. Ongoing efforts to facilitate early recognition of sepsis among patients, caregivers, and clinicians are clearly justified based on these data and should help to chip away at early sepsis-related mortality. Embedding a systematic screening algorithm within clinical workflow, as recently recommended by the Surviving Sepsis Campaign (7), is one tool that has shown promising benefits (8–10). Consistent, goal-directed application of antimicrobial and cardiopulmonary therapies will also be necessary to reduce early sepsis-related deaths in children that largely involve shock progressing to cardiac arrest.

However, the persistence of early deaths in children with sepsis despite two decades of improved recognition, application of aggressive, protocol-driven resuscitation, and emphasis on timely antimicrobial therapy also highlights the need to better understand why, despite optimal care, some children still die early and how to prevent such deaths. The only way to answer these questions is to ensure that children at risk for early death are included in observational and interventional trials. Dr. Johnson and colleagues rightly point out that exclusion of early deaths will likely lead to selection bias. Unfortunately, patients who die early are often not included in such studies because they are not alive long enough to complete informed consent and randomization procedures, and it is challenging to approach the clinical team and family in the throes of active, ongoing resuscitation (which often seems to be in the middle of the night when research staff are less available). For example, despite proactive attempts to enroll children with meningococcal sepsis as early as possible in a study of recombinant bactericidal/permeability-increasing protein, logistical challenges led to 57 children excluded because of early death prior to enrollment and an additional 18 dying within hours of starting study drug (11).

Strategies that facilitate access to both observational and interventional studies for children with sepsis at risk for early death are critical. Ensuring that the sickest children are not inadvertently excluded from research opportunities is part of the ethical principle of justice. Such efforts should include enhanced partnerships between critical care, emergency, and pre-hospital providers in order to screen, enroll, and begin studies prior to PICU arrival. Recognizing that many critically ill children initially seek care at community-based hospitals also underscores the need to involve a broad network of hospitals and transport systems in the research process (when safe and appropriate to do so), rather than confining research exclusively at academic centers. When prospective informed consent is not feasible, proper use of waiver of, exception from, and delayed informed consent should be employed (12–14). Research protocols can also be embedded within sepsis protocols and order sets so that biological samples are collected as early as possible; consent to include the samples in research can then be sought at a later time. Such creative solutions can be both effective and ethical, especially when designed alongside early consultation with an Institutional Review Board. Indeed, several recently completed or ongoing studies demonstrate effective strategies. For example, the FISH (15), SQUEEZE (clinicaltrials.gov/NCT03080038), and PRoMPT BOLUS (clinicaltrials.gov/NCT04102371) interventional studies all recruit patients in the pre-ICU setting under an exception from or delayed informed consent process.

The study by Dr. Johnson and colleagues reminds us that sepsis can still be a rapidly lethal disorder. Our mission, of course, is a tireless commitment to zero sepsis-related deaths. Until that day is here, including children who tragically die early from sepsis today in research studies can be part of their legacy towards a better tomorrow.