Healthcare-associated infections (HAIs) are common in critically ill children and are frequently associated with indwelling devices such as endotracheal tubes, urinary catheters, and central-venous catheters (1–3). Studies suggest these infections are associated with poor outcomes (4). Delays in recognition of HAIs and administration of appropriate antibiotics may further worsen outcomes although data are mostly extrapolated from septic shock or adult data (5,6). Understandably, fear and liability of missing such a serious infection can lead to over-testing, specifically obtaining microbiologic cultures (7). However, despite cultures being a gold standard for diagnosing bacterial infections, results sometimes represent contamination or colonization rather than true infection (8,9). Clinicians may be tempted to initiate empiric antibiotics whenever cultures are obtained and, despite knowing that microbiologic results could be flawed, may continue multiple antibiotics, especially in patients who remain critically ill or when symptoms improve with antibiotics. This can lead to repetitive cycles of fever, cultures, and antibiotics without other clinical evidence of clear infection. Hence, when cultures are frequently obtained without careful consideration, false-positive rates can increase, and unnecessary antibiotic exposures occur, subsequently triggering adverse effects, antibiotic resistance, longer hospital stay, and increases in costs.

In this issue of Pediatric Critical Care Medicine, Sick-Samuels et al (10) report results of their attempts to address the concern of culture overutilization through diagnostic stewardship. Although antibiotic stewardship might be more familiar to many, diagnostic stewardship can fill a similar role by improving efficiency of diagnostic testing and treatment decisions through reducing unnecessary or inappropriate testing (9,11). This group has previously published studies focusing on diagnostic stewardship for PICU patients, including studies from the BrighT STAR collaborative. These studies, both single-center and multicenter, have examined strategies to reduce unnecessary blood or endotracheal aspirate cultures through algorithms that support decision-making in ordering cultures for HAIs (12–14). This current single-center quality improvement project is unique because, based on previous experiences (12–15), Sick-Samuels et al (10) developed and implemented a comprehensive algorithm that uses clinical data to guide risk-based ordering of cultures including blood, endotracheal aspirate, and urine in patients with new concern for HAIs. The algorithm not only helps identify patients with higher pretest probability for HAIs, but also lists other differential diagnoses for fever to consider, something frequently missing when clinicians fall victim to the assumption that all fevers are due to infection rather than considering alternatives.

Sick-Samuels et al (10) included PICU and pediatric cardiac ICU (PCICU) patients and compared the rate of blood, urine, and tracheal aspirate cultures, urinalysis, and broad-spectrum antibiotics ordered during 24 months preimplementation of this expanded algorithm (preintervention) with 18 months postimplementation (postintervention). Although the blood and endotracheal culture rates had already decreased with previous efforts (12–14) and median central-venous catheter-days and ventilator-days were higher in the postintervention group, comparing 4,290 preintervention patients to 2,843 postintervention, blood and endotracheal culture rates further declined by 17% and 26%, respectively. Urine culture rates also declined by 36% and urinalysis rates increased by 23%, which were likely expected since their updated algorithm added a focus on urinary tract infections. In addition, antibiotic initiations also decreased by 12%. Notably, blood and urine culture rates decreased after the third algorithm revision that incorporated feedback from the PCICU team. This likely reflects the effort by Sick-Samuels et al (10) in constantly seeking feedback, fine-tuning the algorithm, and performing frequent walk rounds and surveys to achieve better compliance and buy-in. Balancing measures of in-hospital mortality, 7-day readmissions, and PICU length of stay (LOS) did not increase and there were no concerns of mortality potentially related to algorithm implementation. Accordingly, they demonstrated that a comprehensive algorithm could safely decrease culture utilization.

The determination by Sick-Samuels et al (10) to improve such complex issue is praiseworthy. Still, there are a few points worth discussing. First, since the study does not report how many patients were eligible with new fever or instability during each period, it is possible that there were fewer opportunities to even use the algorithm during postintervention, causing a spontaneous decrease in overall culture rates without any actual change in behavior. Second, it is not clear that there was a stable baseline rate of blood cultures and central line blood cultures during the preintervention period. Several values above the upper control limit may have raised the baseline mean above what was the true baseline culture rate. It may not be fair to compare postintervention data with preintervention data that includes several outliers. Also, the median hospital LOS increased postintervention, something attributed to higher illness severity in the postintervention group, but which could be an unintended consequence of the intervention as well. These consequences may include delaying discharges because of fear of missing HAIs without culture results or prolonged hospital stay because of the impact of delayed antibiotic initiation. Furthermore, while they did not report mortality related to missed infections, the incidence of delayed diagnosis and treatment of HAIs was not reported. There were likely febrile patients that were not cultured based on the algorithm, only to have an HAI diagnosed later, contributing to longer hospital LOS. In addition, compliance with and acceptance of the algorithm is unclear. Based on the survey results, only 36% reported using the algorithm (39 of 108 physicians) and 26% (28 physicians) reported improvement in care. The possibility of low buy-in, including other services, cannot be excluded. This could pose challenges when other units, with a potentially less motivated group of clinicians, attempt to implement this approach in their PICUs.

A possible larger concern is that the broad-spectrum antibiotic days of therapy (DOT) did not change based on the algorithm. This may be problematic since cumulative antibiotic exposure likely has more negative consequences than the number of times antibiotics are started and then discontinued. Although likely multifactorial, one should question whether DOT did not change because clinicians could not discontinue or de-escalate antibiotics without appropriate cultures. This is particularly possible in ventilator-associated infection (VAI) because the algorithm suggests not collecting lower respiratory cultures if one was obtained within the prior three days, regardless of whether there were new symptoms or if broad-spectrum antibiotics were being initiated. Clinical changes commonly associated with VAI (e.g., color or consistency of secretions) were considered irrelevant findings. Hence, despite potential concern for new VAI, the algorithm may have prompted treatment of VAI without microbiological data, leading to prolonged antibiotic courses without de-escalation. Clinicians may also opt to treat empirically without obtaining cultures due to fear that they might get in trouble for not following the algorithm. Regardless of the reason, no change in DOT suggests potential minimal improvement in downstream detrimental effects related to antibiotics including side effects and resistance, and might explain the lack of impact on LOS.

The study by Sick-Samuels et al (10) represents a great example of multidisciplinary diagnostic stewardship. Decreasing excessive testing is challenging, especially in detrimental diseases such as sepsis and HAIs. However, given the reluctance of many intensivists to not treat bacteria with antibiotics when they are identified, regardless of whether other signs of infection are present, reducing unnecessary microbiologic testing seems to be a way forward to improve antibiotic utilization and infection management. The challenge one will most likely face when implementing a similar algorithm is practitioner inertia, from fear of missing diagnosis or disagreement either with the need for or approach of the algorithm. However, as novel diagnostic tests with high sensitivity such as blood microbial cell-free DNA and multiplex polymerase chain reaction for pneumonia become more readily available, the complexity of distinguishing true bacterial infection from colonization or false-positives will increase, further encouraging need of diagnostic stewardship. As the study by Sick-Samuels et al (10) exemplifies, a diligent yet flexible approach reflecting institution-unique needs will be crucial to change the PICU culture surrounding fever and diagnostic testing.

1. Hsu HE, Mathew R, Wang R, et al.: Health care-associated infections among critically ill children in the US, 2013-2018. JAMA Pediatr. 2020; 174:1176–1183 2. Foglia E, Meier MD, Elward A: Ventilator-associated pneumonia in neonatal and pediatric intensive care unit patients. Clin Microbiol Rev. 2007; 20:409–425 3. Patrick SW, Kawai AT, Kleinman K, et al.: Health care-associated infections among critically ill children in the US, 2007-2012. Pediatrics. 2014; 134:705–712 4. Kollef MH, Torres A, Shorr AF, et al.: Nosocomial infection. Crit Care Med. 2021; 49:169–187 5. Weiss SL, Fitzgerald JC, Balamuth F, et al.: Delayed antimicrobial therapy increases mortality and organ dysfunction duration in pediatric sepsis. Crit Care Med. 2014; 42:2409–2417 6. Zilberberg MD, Shorr AF, Micek ST, et al.: Antimicrobial therapy escalation and hospital mortality among patients with health-care-associated pneumonia: A single-center experience. Chest. 2008; 134:963–968 7. Woods-Hill CZ, Koontz DW, King AF, et al.; Bright Star Authorship group: Practices, perceptions, and attitudes in the evaluation of critically ill children for bacteremia: A national survey. Pediatr Crit Care Med. 2020; 21:e23–e29 8. Willson DF, Conaway M, Kelly R, et al.: The lack of specificity of tracheal aspirates in the diagnosis of pulmonary infection in intubated children. Pediatr Crit Care Med. 2014; 15:299–305 9. Fatemi Y, Bergl PA: Diagnostic stewardship: Appropriate testing and judicious treatments. Crit Care Clin. 2022; 38:69–87 10. Sick-Samuels AC, Booth LD, Milstone AM, et al.: A Novel Comprehensive Algorithm for Evaluation of PICU Patients With New Fever or Instability. Pediatr Crit Care Med. 2023; 24670–680 11. Sick-Samuels AC, Woods-Hill C: Diagnostic stewardship in the pediatric intensive care unit. Infect Dis Clin North Am. 2022; 36:203–218 12. Woods-Hill CZ, Fackler J, Nelson McMillan K, et al.: Association of a clinical practice guideline with blood culture use in critically ill children. JAMA Pediatr. 2017; 171:157–164 13. Woods-Hill CZ, Colantuoni EA, Koontz DW, et al.; Bright STAR Authorship Group: Association of diagnostic stewardship for blood cultures in critically ill children with culture rates, antibiotic use, and patient outcomes: Results of the Bright STAR collaborative. JAMA Pediatr. 2022; 176:690–698 14. Sick-Samuels AC, Linz M, Bergmann J, et al.: Diagnostic stewardship of endotracheal aspirate cultures in a PICU. Pediatrics. 2021; 147:e20201634 15. Booth LD, Sick-Samuels AC, Milstone AM, et al.: Culture ordering for patients with new-onset fever: A survey of pediatric intensive care unit clinician practices. Pediatr Qual Saf. 2021; 6:e463