Clinical and molecular markers to assist decision-making in neonatal sepsis

Introduction

Neonatal sepsis is characterized by non-specific signs and difficulty in distinguishing it from non-infectious conditions, but also by a rapid progression toward dysfunction of vital organs, high mortality, and a risk of lifelong disability, especially in preterm infants. Therefore, clinicians have a low threshold to initiate empirical antibiotics. In fact, suspected neonatal infection or “rule-out sepsis” is one of the most common diagnoses in neonatal units, and antibiotics are among the most frequently prescribed medications. However, in the vast majority of cases, cultures remain negative, and sepsis is subsequently ruled out. Therefore, the current approach leads to substantial overtreatment with antibiotics, and exposes neonates to colonization with antibiotic resistant bacteria. Antibiotic treatment also induces perturbations in early-life microbiota with potentially long-lasting negative impact on the individual’s health. Accurate diagnostic algorithms and tools are needed to accelerate the diagnosis of neonatal sepsis while minimizing overtreatment. Currently, most published studies analyzing risk factors, clinical signs, and biomarkers/molecular markers have major limitations regarding design (observational studies), size (insufficient power to prove safety), and strategy (lack of acknowledging important factors such as baseline sepsis incidence or focusing on a single or a limited set of clinical or molecular markers).

This chapter presents the current state of knowledge on the use of clinical and molecular markers to guide the management of neonatal sepsis, based on data from high-income countries. The diagnostic value of and potential pitfalls in the interpretation of clinical and laboratory data to guide decisions to start, stop, or continue antimicrobial treatment and for risk stratification are reviewed. Moreover, we discuss how clinical and molecular markers can be integrated in a framework aiming at optimizing the decision-making process for suspected neonatal sepsis.

Decision-making in neonatal sepsis

Decision-making is the backbone of clinical medicine. Despite research and guidelines, an understanding of how decisions are made in daily clinical practice remains elusive. Very large variations in the use of antibiotics among neonatal units have been described. To some extent, they can be explained by differences in guidelines, patient populations, level of implementation of antimicrobial stewardship programs, and leadership style at each institution. However, it is more difficult to assess the impact of seniority, interactions between junior and senior physicians, habits, and dramatic or negative experiences on decision-making by individual clinicians and teams. Excessive fear of sepsis may be a key driver for antibiotic overtreatment in neonates. Whereas a high face validity exists for this assumption, literature is scarce regarding this topic. There is some evidence from qualitative studies that fear critically influences antibiotic decision-making. A systematic review of qualitative studies regarding physician antibiotic-prescribing behavior found that complacency and fear were the most influential intrinsic factors. Moreover, when faced with similar scenarios, individual clinicians may not always make the same decisions. Decision-making in neonatal sepsis is affected by bias, a systematic deviation toward overtreatment, and noise, a random scatter.

The decision to start antibiotics is typically based on a prediction made by the clinician that a particular patient has an ongoing bacterial infection. Some predictive judgments are verifiable, but this is rarely the case in neonatal sepsis. One can only be certain that starting antibiotics was the correct decision if the patient has an unambiguous proof of infection, documented by a positive and non-contaminated blood or cerebrospinal fluid culture. As this is rarely the case in clinical practice, many decisions regarding initiation and termination of antibiotics are made under conditions of uncertainty. The quality of unverifiable judgments can only be assessed by the quality of the thought process that produces them.

Cognitive biases in the interpretation of clinical scenarios (including data from clinical and molecular markers of sepsis) contribute to inadequate antibiotic prescribing practices. However, the capacity of clinicians to make predictions might be greater than expected. In a prospective study of 347 patients admitted to a neonatal and a pediatric intensive care unit, physicians were asked, based on all available information, to provide an estimate of the presence of serious infection at every ward round. In cases for which empirical antibiotics were started, predictions made at initiation of antibiotics discriminated well patients who were later deemed to have culture-proven sepsis from non-infectious episodes (area under the receiver operating characteristic curve [AUROC] = 0.88). But, only 14% of antibiotic courses were administered for episodes of proven systemic bacterial infection, suggesting that accurate predictions do not always translate into appropriate clinical decisions. Clinical and molecular markers should not be considered as single measurements allowing to rule in or rule out sepsis on their own. They should be interpreted building on evidence-based data on their diagnostic value and their potential to add value to the judgment of physicians and improve patient outcomes as part of an overall structured approach to optimize antibiotic use in early life. As an example, the Agency for Healthcare Research and Quality provides a program to improve antibiotic-prescribing practices by combining adaptive change theories and evidence-based diagnostic and treatment practices. Recently, an approach based on six key questions to evaluate the objective risk of bacterial infection was proposed to guide antibiotic prescriptions in critically ill children, and overcome irrational decision-making due to the fear of missing severe bacterial infections.

Conceptual framework for decision-making in neonatal sepsis

Key aspects in the decision-making process are the determination of the probability of sepsis and the evaluation of the risk–benefit ratio of starting, not starting, or stopping antibiotics for a given patient. We provide a conceptual framework to assist clinicians in decision-making for suspected neonatal sepsis ( Figure 19.1 ). Assessment of the probability of sepsis in a given patient can be done using a structured approach, including the following steps: (1) determination of baseline incidence of sepsis, (2) evaluation of risk factors and clinical and molecular markers and consideration of alternative diagnoses to estimate the specific probability of sepsis for an individual patient, (3) evaluation of the course of clinical and molecular markers to adapt the assessment of the specific probability of sepsis, and (4) proof or definitive exclusion of sepsis with the help of microbiology results. Importantly, the course of clinical signs and serial evaluation of molecular markers will provide more information than a single assessment at the onset of symptoms. Guidelines and educational programs are key components of the decision-making framework and may assist clinicians at every step of the process.