Authors’ reply

We thank L Nelson Sanchez-Pinto and colleagues for their comments on our Correspondence¹ and concur that every avoidable death of a child, or adult, is a death too many. However, their comments underline our points precisely—namely, issues with definition, denominators, attributable deaths, and avoidable deaths due to lack of timely antibiotics.

We focused on data from the UK and the USA, where mortality associated with sepsis predominates in older people (aged ≥75 years). Data from the National Health Service England identified approximately 1·1 million emergency hospital admissions in people aged 0–19 years between 2011 and 2017 with a discharge code of sepsis or infection, yet only 892 deaths (<1%). Thankfully, paediatric sepsis trials in high-income countries have not used mortality as the primary endpoint for many years, citing “affordability”.² A point-prevalence study of 6925 children across 128 paediatric intensive care units in 26 countries identified 569 children who fulfilled sepsis criteria.³ Of these patients, 439 children (77%) had one or more comorbid conditions that could independently limit survival, including solid organ or stem-cell transplants, malignancy, renal disease, and haematological or immunological conditions. Fungal and viral pathogens were frequently identified in these patients as causative. The authors estimated that an intervention trial would take 3 years for 2118 patients to be enrolled from 58 paediatric intensive care units for 80% power to detect a 5% absolute risk reduction in mortality. This figure equates to an enrolment rate of one patient per month.

With respect to low-income and middle-income countries, paediatric mortality rates are indeed higher than in high-income countries; however, what is the true incidence, and in what proportion could antibacterial therapy be potentially life-saving? The 32% case fatality rate reported by Tan and colleagues⁴ relates only to the severe end of the sepsis spectrum, but they also showed a progressive fall in case fatality rate from 55% in 1996 to 25% in 2016. Two major prospective case-control studies on children hospitalised in sub-Saharan Africa and Asia with moderate-to-severe diarrhoeal disease (9439 participants)⁵ or severe pneumonia (4232 participants)⁶ also offer a different perspective. Among diarrhoeal illnesses, most cases were attributable to viral or parasitic causes, with an overall 90 day mortality rate of 2%. In the study on severe pneumonia, 30 day mortality was 7·3%, with viral pathogens predominating. Even in the cohort of patients with very severe pneumonia, bacterial pathogens (predominantly Streptococcus pneumoniae and Haemophilus influenzae type B) only accounted for a third of cases. Mortality rates are likely to be more affected by increasing availability of vaccinations and health-care resources, and improving sanitation and nutrition, than by antibacterials.

Our plea was to contextualise sepsis appropriately and to deliver accurate sepsis epidemiology rather than estimates that might greatly misrepresent the truth (in either direction), and to adopt a balanced approach to antibiotic prescribing, media reporting, and public education. Most children with infections will not benefit from antibiotics and might be harmed by the effects on microbiota and other unwanted complications, notwithstanding the impact on antimicrobial resistance.