Prospective surveillance with intervention and feedback is a core strategy of effective antibiotic stewardship, but it has not been applied rigorously to antibiotic use in the NICU. Despite a growing body of literature on adverse effects associated with prolonged or unnecessary antibiotic therapy in the NICU, there has been little focus on why and how often these antibiotics are used. To our knowledge, this study represents the most comprehensive prospective evaluation of antibiotic use in the NICU. We identified several areas that should inform stewardship efforts and highlight clinical scenarios where further study is needed.
Only a small fraction of antibiotic use in the NICU was directed toward proven infection. Culture-proven sepsis, necrotizing enterocolitis, cellulitis and congenital infections combined to account for <7% of all antibiotic use. Pneumonia represented an additional 16% of all antibiotic use, although a portion of these infants may have had noninfectious conditions such as transient tachypnea of the newborn or respiratory distress syndrome rather than pneumonia.[31,32] The largest amount of antibiotic therapy was empiric therapy for suspected sepsis (Fig. 2). Accurate identification of septic infants in the NICU remains a challenge, as neonates can present with nonspecific findings that overlap with noninfectious etiologies.[33,34] Furthermore, at present, there is no combination of clinical findings or laboratory testing that has sufficient sensitivity to preclude the need for empiric therapy. Therefore, stewardship efforts must focus on timely discontinuation of antibiotic therapy once infection is no longer suspected. We observed a high rate of inadvertent continuation of antibiotic therapy beyond 48 hours despite clinicians documenting intent to stop therapy within 48 hours (68% of "ruled-out sepsis" courses). These extra doses accounted for an additional 40.1 DOT/1000 PD (22.3 LOT/1000 PD), or 12% of the total antibiotic use during the study period. A "hard stop," or automatic discontinuation of empiric therapy at 48 hours by the computerized physician ordering system, may be an effective means of reducing inadvertent continuation of antibiotics, but will require close safety monitoring to prevent inadvertent discontinuation of needed therapy.
Pneumonia accounted for 16% of all antibiotic use during the study period, and it was the most common indication for prolonged therapy. There was significant variation in the treatment duration of infants diagnosed with pneumonia. The median duration of therapy was 7 days; approximately two thirds of infants received ≥7 DOT, while one third received ≤5 days. Engle et al performed a small, randomized trial for term infants in the newborn nursery diagnosed with pneumonia. Those who were asymptomatic by 48 hours were randomized to either 4 or 7 days of antibiotic therapy. No difference in short-term outcomes was found. Though small and not generalizable to the NICU, this study provides a basis for further research and validation. Although the diagnosis of neonatal pneumonia remains problematic and challenging, larger clinical trials evaluating the safe minimum duration of therapy for pneumonia could have a significant impact on antibiotic stewardship strategies.
"Culture-negative" sepsis was the second most common indication for prolonged therapy. Similar to pneumonia, approximately two thirds of infants received ≥7 DOT, while one third received 5 days. Suspected sepsis remains a challenge in the NICU. The clinical signs that result in the initiation of a sepsis evaluation may be due to problems associated with prematurity rather than infection, and distinguishing the 2 often is difficult.[32,38,39] Infants who do not have another explanation for their illness may receive prolonged therapy for suspected sepsis, particularly when cultures are not obtained or are obtained after antibiotic therapy has been initiated. Debate continues on the optimal amount of blood for culture, and whether the lack of a sufficient quantity may preclude identification of a bacterial cause. The sensitivity of blood cultures is not 100%, although it approaches 100% at adequate volumes (≥1 mL). Furthermore, infants born to mothers who have received intrapartum antibiotic prophylaxis may be viewed as having falsely negative blood cultures, although this is counterintuitive as intrapartum antibiotic prophylaxis is associated with a lower risk for sepsis. Similarly, abnormal white blood cell counts are common among infants born to mothers with chorioamnionitis and antibiotic therapy may be discontinued safely in these infants if they remain clinically well.[43,44] Studies of both preterm and term infants, however, suggest that clinicians may not base duration of therapy for suspected sepsis on risk factors or physical examination findings, but rather on subjective factors.[45,46] To minimize antibiotic exposure, it is imperative that providers obtain appropriate cultures of all body sites (eg, blood, urine, cerebrospinal fluid), and then trust those results.
Antibiotic stewardship efforts often focus on broad-spectrum antibiotic use. Preauthorization is a key aspect of stewardship and is intended to steer clinicians toward narrow-spectrum antibiotic therapy. However, over 92% of antibiotic use in our study was narrower spectrum therapy (ampicillin, oxacillin, gentamicin and penicillin) that is not monitored or restricted routinely by antibiotic stewardship programs.[48,49] Less is known about the potential adverse effects of such narrower spectrum therapy. Although these agents may be viewed as safe by many providers, they may alter the infant's microbiome and are associated with an increased risk of necrotizing enterocolitis or death in very preterm infants.[6,7] Further study is needed to determine the short-term and long-term impact that these agents may have on vulnerable neonates.
Our study has several limitations. The optimal metric for antibiotic use in the NICU has not been defined and requires further study.[27,49,51,52] It is not known which metric best represents the degree of antibiotic exposure and thus the impact on the infant's microbiome. For these reasons, we chose to report both DOT and LOT, although most pediatric antibiotic stewardship efforts use DOT as the main benchmark.[25,52] In addition, there was a risk for observer effect in the prospective portion of the study, as the neonatologists and fellows were aware of the surveillance. However, there was no significant change in the frequency of antibiotic use between the retrospective period and prospective period, suggesting that any observer effect was minimal. EMR review was performed by 1 study member (J.B.C.), so interrater reliability of the reasons for antibiotic use could not be examined. Finally, our study may not be generalizable to other NICUs, especially those in children's hospitals where most or all infants are outborn or admitted from the community. Surveillance can guide institution-specific interventions.
In conclusion, prolonged or excessive antibiotic therapy has been associated with a variety of adverse outcomes.[3–10] Stewardship efforts to minimize unnecessary therapy are needed urgently, but must be informed by appropriate prospective surveillance of antibiotic use. "Ruled-out" sepsis evaluations, pneumonia and "culture-negative" sepsis courses are high-yield targets for antibiotic stewardship interventions; culture-proven infections account for a small fraction of antibiotic use. Focusing on timely discontinuation of therapy once infection is ruled out as well as evaluating the safe, minimum duration of therapy for common clinical scenarios will help to reduce antibiotic use in the NICU and avoid the adverse outcomes associated with their use. While routine prospective surveillance is time-consuming and may not be practical for all centers, prospective audit with feedback and interventions is a keystone of good antibiotic stewardship and should be a focus in the NICU.
The authors thank John Gard, PharmD, Melody Bush, RPh, Sara Mureeba, PharmD and Sheeba Tharayil, PharmD, of the PMH NICU pharmacy for their help with the study. Dr. Cantey had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Pediatr Infect Dis J. 2015;34(3):267-272. © 2015 Lippincott Williams & Wilkins