Timing of Death in Children Referred for Intensive Care With Severe Sepsis: Implications for Interventional Studies

Objective: Early deaths in pediatric sepsis may limit the impact of therapies that can only be provided on PICUs. By introducing selection and survivorship biases, these very early deaths may also undermine the results of trials that employ standard consent procedures. We hypothesized that: 1) the majority of deaths in children with severe sepsis occur very early, within 24 hours of referral to PICU; and 2) a significant proportion of deaths occur before PICU admission. Design, Setting, and Patients: We studied consecutive referrals of newborns through to 16 years of age, between 2005 and 2011 to the Children's Acute Transport Service, the North Thames regional pediatric intensive care transport service, with a working diagnosis of sepsis, severe sepsis, meningococcal sepsis, or septic shock. Interventions: The primary outcome measure was the proportion of deaths within 24 hours of referral. Survival distributions of previously healthy children were compared with those with significant comorbidities. Measurements and Main Results: Thirteen thousand four hundred and nine referrals were made to Children's Acute Transport Service, of whom 703 (5%) met inclusion criteria. Data on survival to 1 year were available in 627 of 703 patients (89%). One hundred thirty children (130/627; 21%; 95% CI, 18-24%) died in the first year. A higher proportion of children with comorbidity cases (46/85, 54%, 44-64) died compared with previously healthy cases (84/542; 16%; 13-19; p < 0.0005, Fisher exact test). Seventy-one deaths occurred within 24 hours of PICU referral (71/130, 55%, 46-63). The timing of death differed with comorbidity. Similar proportions of children survived to 24 hours (previously healthy children 90% vs children with comorbidity 83%, p = 0.06). However, deaths after 24 hours were infrequent among previously healthy cases (28/84 deaths, 33%, 24-44%) compared with children with comorbidity cases (31/46 deaths, 66%, 53-79%) (p < 0.001, Fisher exact test). Conclusions: This majority of deaths among children referred for pediatric intensive care with for severe sepsis occur within 24 hours. This has important implications for future clinical trials and quality improvement initiatives aimed at improving sepsis outcomes.