Disparities in the Care of Febrile Infants: Embedding Equity in Every Step of Quality Improvement

The evaluation and management of young febrile infants have evolved over several decades. Most recently, the American Academy of Pediatrics (AAP) published clinical practice guidelines for the evaluation and management of well-appearing term infants aged 8 to 60 days, with the intent of providing evidence-based recommendations that account for changing bacteriology and promote high value care via reductions in invasive testing, unnecessary antimicrobial administration, and costly hospitalizations.1 The guidelines, which include 21 key action statements, underscore the utility of inflammatory markers (IMs) like procalcitonin (PCT), which is highly specific for invasive bacterial infections (IBIs), in the risk stratification of febrile infants and to inform decisions regarding performing lumbar punctures, initiation of antimicrobials, and hospitalization. The authors of the AAP guidelines acknowledge the potential impact of variable access to newer diagnostic tests, particularly PCT, and timely results on successful guideline adherence and implementation. Additionally, they underscore that even in settings in which ready access to such testing exists, the variable levels of risk tolerance among both clinicians and caregivers, which are informed by several factors, likely influence guideline interpretation and adherence.In this issue of Hospital Pediatrics, Kelly et al describe a quality improvement (QI) initiative spurred by low rates of PCT testing and high rates of hospitalization and antibiotic administration among febrile infants aged 22 to 56 days evaluated in their emergency department (ED).2 They aimed to safely decrease the proportion of these infants who received intravenous antibiotics from a baseline of 46% to 36% in 1 year, inclusive of race and ethnicity and preferred language. First, they formed a multidisciplinary team to develop change ideas related to the articulated aim. Second, they completed 4 Plan-Do-Study-Act (PDSA) cycles—all centered on clinical pathway updates and corresponding provider education and electronic order set adaptations. More specifically, the team (1) modified the clinical pathway to include use of IMs (PCT, C-reactive protein [CRP], and absolute neutrophil count [ANC]); (2) transitioned from use of an enhanced urinalysis to a standard urinalysis; (3) removed CRP as a recommended IM; and (4) removed neutropenia (ANC < 1000) as an abnormal IM. They developed clear operational definitions for their (1) primary measure—proportion of eligible infants receiving at least 1 dose of an intravenous antibiotic in the ED; (2) process measure—proportion of infants with a PCT result; and (3) balancing measure—missed IBI. The team tracked measures over time via statistical process control charts and then disaggregated the primary and process measures by race and ethnicity and preferred language. They achieved an overall decrease in antibiotic use among febrile infants aged 22 to 56 days from 46% to 33% within 1 year; infants 22 to 28 days old had the greatest decrease in antibiotic administration from 86% at baseline to 43% after intervention. The authors also observed a marked increase in the proportion of febrile infants with a PCT result, from 4% at baseline to 97% after intervention.Notably, a stated goal of the improvement effort was to pursue the aim in a manner “inclusive of race/ethnicity and preferred language.” The disaggregation and tracking of measures of interest by sociodemographic characteristics like race and ethnicity and preferred language should be the standard for all QI efforts, and the power of doing so is illustrated in this work. Kelly and colleagues did not observe disparities in the proportion of infants with a PCT result across different racial and ethnic and preferred language subpopulations. However, although the baseline antibiotic administration rates among non-Hispanic white infants and non-Hispanic Black infants were similar (45% vs 47%, respectively), a significant decrease in antibiotic administration rates after intervention was only observed among non-Hispanic white infants compared with non-Hispanic Black infants (29% vs 47%, respectively), thus illustrating a widening equity gap. Likewise, baseline antibiotic administration rates among infants of families with a preferred language of English and infants of families with a preferred language other than English were similar (47% vs 50%, respectively), but a significant decrease in antibiotic administration was only observed among infants of families with a preferred language of English compared with infants of families with a preferred language other than English (33% vs 50% after intervention, respectively). It is important to note that the authors included all febrile infants in their denominator and did not further stratify the data of infants who received antibiotics by low (<0.5) vs high (≥0.5) PCT, which limits the ability to interpret the findings. However, given the low prevalence of IBIs in this population, the findings may indicate that although clinicians obtained PCT levels reliably among nearly all febrile infants, the impact of PCT values on their clinical decision-making—even when reassuring—likely varied across demographic groups.Teams within health care systems should incorporate key principles in their pursuit of excellent and equitable outcomes when undertaking QI work.3–5 Equity-oriented QI must begin by laying the groundwork before active testing of interventions takes place. This requires an intentional focus on defining aims that not only name subpopulations of interest but also proactively identify problems and failure modes that may disproportionately impact them. Teams should determine at the outset how to amplify the perspectives of individuals with lived experience by inviting them to be a part of the multidisciplinary QI team and leverage other coproduction tactics. Furthermore, key drivers, or “the what” that needs to be in place to achieve the stated goal, must be inclusive of drivers that address identified or potential equity gaps. The authors prioritized PDSAs solely focused on clinical pathway adaptation and implementation. However, articulated drivers to address specific factors that are likely to contribute to disparities—such as practice variation due to differences in risk tolerance and clinician bias—were not a part of the work. The authors also do not mention the inclusion of patients or families on the improvement team, particularly those from historically marginalized backgrounds.Additionally, equity-oriented measurement approaches are foundational to improvement work. Teams should define the sociodemographic variables they will use to segment the data and share relevant information about data collection methods, quality, and rates of missing data. This is of particular importance for race, ethnicity, and language variables. Teams should consider the inclusion of other relevant sociodemographic variables such as insurance type and geography. Disaggregated data must be proactively tracked in real time as the improvement work ensues to allow for timely mitigation and testing of interventions to directly address equity gaps. Retroactive identification of equity gaps, although still important, is a striking depiction of missed opportunities over time. In this case, non-Hispanic Black infants and infants of families with a preferred language other than English have continued to be subjected to unnecessary lumbar puncture, antibiotics, and/or hospitalizations at disproportional rates.Pediatric health inequities are common and disproportionately impact poor children and racial and ethnic minorities; racism and discrimination in all its forms, including implicit bias, are root causes.6 Several studies have described racial disparities in the odds or likelihood of ordered laboratory and radiology studies for children who present to the ED, including young febrile infants.7,8 For example, an analysis of febrile infants aged 29 to 60 days deemed to be low risk of IBIs based on blood and urine testing showed that infants of families with a preferred language other than English had higher odds of receiving at least 1 additional intervention (ie, lumbar puncture, empirical antibiotics, or hospitalization) contrary to evidence-based recomendations.9 Additionally, several disparities were noted in a recent cross-sectional study of infants aged 8 to 60 days across 99 US hospitals engaged in a QI collaborative to support implementation of the 2021 AAP guidelines, including disparities in appropriate IM obtainment, documentation of shared decision-making with families regarding lumbar punctures, and documented follow-up from the ED.10 These findings illustrated the unintended consequences of standardization and improvement efforts when equity is an afterthought and not actively pursued.The 2021 AAP guidelines delineate opportunities for clinician discretion and shared decision-making between clinicians and families, particularly for infants aged 22 to 28 days with reassuring blood and urine testing.1 Shared decision-making can improve patient and family experience, although notably, racial and ethnic minority children and families as well as those with a preferred language other than English have fewer opportunities to engage in shared decision-making and experience differences in the degree of emotional support and information exchange during the process.11,12 Thus, although clinician discretion and shared decision-making may provide opportunities for individualized, patient- and family-centered care, the intrusion of bias is a harsh reality.Kelly and colleagues demonstrate the power of QI methodology in facilitating guideline adherence and standardizing diagnostic practices for febrile infants who present to the ED. However, the notable disparities in antibiotic administration among racial, ethnic, and preferred language subpopulations underscores the need for a proactive and intentional approach to embed equity within every step of QI efforts. To accomplish this, teams and institutions should (1) include individuals with lived experience as partners in the work; (2) define aims and invest time outlining current state and failure modes that may disproportionately impact subpopulations of interest; (3) derive equity-oriented key drivers that will inform interventions; (4) deploy a thoughtful approach to defining, disaggregating, and tracking process, outcome, and balancing measures; and (5) actively test interventions to address equity gaps identified prior to the start of the work and/or as they become apparent. Such approaches will bring us closer to bridging the divide between the care we intend to deliver to all patients and the inequitable care that specific subpopulations of patients receive.