Antimicrobial resistance (AMR) is an urgent threat to global health, accelerated by the inappropriate use of antibiotics. Antimicrobial stewardship (AMS) is crucial for optimizing patient outcomes, minimizing drug toxicity, and the emergence of resistance. Diagnostic stewardship (DS) is an important component of AMS, requiring the right tests to be ordered for the right patients at the right time. It also promotes the use of rapid and novel molecular diagnostic tools to allow for the initiation of appropriate antibiotic therapy while avoiding the overuse of broad-spectrum antibiotics when not needed. Care should be taken when interpreting test results to avoid overdiagnosis and unnecessary costs. Despite good outcomes, rapid diagnostics are limited by cost, accessibility, and misunderstanding. DS relies on thorough history taking and physical examination, targeted diagnostics, appropriate specimen collection, and correct interpretation of results. Additionally, a multidisciplinary team composed of trained professionals should be involved in the diagnostic pathway. Here, we discuss how to implement DS in five common infection syndromes frequently encountered in the Intensive Care Unit (ICU).1. Hospital-Acquired and Ventilator-Associated Pneumonia Ventilator-associated pneumonia (VAP) is a leading cause of mortality in the ICU. However, it is challenging to distinguish between colonization and infection. Furthermore, given the accessibility of respiratory sampling in intubated patients, clinicians tend to request numerous respiratory cultures, potentially misinterpreting them as infections. In fact, up to 50% of patients receiving antibiotics for VAP in the ICU may merely be colonized. On the other hand, delays in antibiotic treatment for VAP are associated with higher mortality rates. Recent studies suggest that microbiological identification prior to treatment initiation in non-septic patients is desirable and may reduce the emergence of resistance. The Infectious Diseases Society of America (IDSA) recommends using clinical and microbiological criteria to diagnose VAP, with reassessment within 48 hours of starting antibiotics. The long turnaround times of traditional methods may delay targeted antimicrobial therapy. A retrospective study conducted in French ICUs during the COVID-19 pandemic utilized a rapid microbiological diagnostic test based on nested multiplex polymerase chain reaction (mPCR) for protected telescoping catheter (PTC) specimens. This semi-quantitative test allows for the detection of 15 bacteria, 3 atypical bacteria, 9 viruses, and 7 antibiotic resistance genes within 1.5 hours of sample collection. The study demonstrated that mPCR performed well on PTC samples, with a sensitivity of 93%, specificity of 99%, and a negative predictive value (NPV) of 100%. In another retrospective multicenter study, conventional microbiological methods and the new syndrome rapid multiplex PCR detection (rmPCR) were used to simultaneously detect respiratory samples. The syndrome rmPCR detected 83% of infections, while the infection rate in conventional cultures was 60%, allowing for treatment escalation/de-escalation in 77% of pneumonia cases. In another study, a multidisciplinary expert panel analyzed 95 samples and simulated the changes that would occur if mPCR were available. Their conclusion was that mPCR improved empirical treatment, reduced the use of broad-spectrum antibiotics, and even diagnosed two unexpected cases of severe Legionnaires’ disease. A multicenter study evaluated two mPCR platforms for rapid microbiological screening of hospital-acquired pneumonia (HAP) in critically ill patients across 15 ICUs in the UK. Both systems were significantly faster than conventional microbiological testing and detected more pathogens than conventional microbiological methods. Importantly, PCR can detect more microorganisms and improve the microbiological diagnosis of pneumonia. Bayesian latent class analysis (BLC) showed that conventional microbiological analysis had lower sensitivity compared to PCR testing, which had higher specificity and positive predictive value (PPV). Furthermore, although PCR does not provide a complete sensitivity profile, it is a rapid predictor of key resistances with infection control significance. Currently, a randomized controlled trial is exploring the potential benefits of mPCR in guiding treatment for ICU patients with HAP/VAP.2. Central Nervous System Infections Central nervous system (CNS) infections are associated with extremely high mortality rates. Clinical presentations often cannot distinguish between bacterial and viral causes. If microbiological identification is delayed, empirical treatment is recommended. Additionally, clinical evaluation may be inconclusive in critically ill patients. Therefore, the use of rapid and accurate diagnostic tools can shorten the time to initiate appropriate treatment and avoid unnecessary antimicrobial use. Microarray PCR testing of cerebrospinal fluid (CSF) is a promising diagnostic tool. It can detect organisms present in low loads, with an accuracy of 90% and a turnaround time of 1 hour. Microarray testing of CSF may be particularly useful in pediatric patients with suspected CNS infections when normal CSF findings are present. Nevertheless, this highly sensitive diagnostic tool may lead to overdiagnosis, especially in cases with low pre-test probabilities. In the absence of guidelines, DS interventions are crucial to help clinicians correctly utilize advanced diagnostic tools to reduce overdiagnosis and unnecessary treatment. Diagnostic algorithms aimed at clinicians have successfully reduced excessive microarray PCR testing. The diagnostic algorithm avoided 75% of false-positive results without generating false negatives. In many cases, the lack of pleocytosis hinders clinicians from performing molecular diagnostics. In fact, except for immunocompromised patients and children under 6 months, the NPV of CSF white blood cells is as high as 98-100%, ruling out CNS infections. Typically, physicians hesitate to under-treat, especially in the ICU, which affects management. One study demonstrated this, where 78% of patients with suspected CNS infections and negative microarray results were treated with antimicrobials. Wise use of high-yield diagnostic tests significantly reduces the time to adequate treatment.3. Clostridioides difficile Infection Clostridioides difficile infection (CDI) is very common among ICU patients receiving antimicrobial therapy. However, up to 50% of nucleic acid amplification test (NAAT) positive patients are colonized rather than infected. The use of highly sensitive NAAT instead of antigen or toxin-based testing increases the risk of overdiagnosing CDI. Furthermore, the lack of specific biomarkers for CDI complicates the diagnosis. “Soft stops” integrated into electronic health systems (EHS), such as reminders to check for laxative use, can facilitate clinical decision-making and improve the appropriateness of testing. Otherwise, strict interventions known as “hard stops,” which block orders without pre-specified criteria, may reduce over-testing by 56%. Additionally, after implementing pre-authorization protocols for CDI testing, a study showed a reduction in oral vancomycin prescriptions. When collecting samples, stool cultures should be collected in clean containers, kept at room temperature, and transported within two hours. Furthermore, microbiology laboratories play a crucial role in facilitating DS. For example, rejecting non-stool samples reduced testing by 43%, and CDI events decreased by 60%. Sample rejection based on pre-specified clinical criteria also helps reduce unnecessary CDI testing without affecting mortality rates. Although negative toxin tests may predict milder disease, clinicians should be aware that toxin-based tests may not have sufficient NPV to rule out CDI.4. Bloodstream Infections Excessive blood culture (BC) ordering is common among ICU patients with vascular and indwelling catheters, who are at high risk for bloodstream infections (BSI). Despite limited correlation, BC is often driven by leukocytosis and fever. In fact, a recent review reported that up to 20% of positive BCs may be contaminated. Contaminated BCs increase exposure to antimicrobials, costs, and length of hospital stay. BC should not be obtained when it is unlikely to change the management of patients with clear infection sites and no sepsis or septic shock. However, BC may be of great value in the presence of syndromes likely to yield positive BCs, such as CNS infections, infectious arthritis, and intravascular infections, or when sampling the primary site of infection is difficult. The DISTRIBUTE study aimed to optimize BC practices by implementing evidence-based BC indication algorithms and providing education and feedback to providers regarding BC rates and appropriateness. It demonstrated that these interventions effectively and safely reduced unnecessary BCs. A machine learning model used in a multicenter validation prospective study indicated that it could safely retain BC analysis in at least 30% of emergency patients. BC sampling should be performed before the use of antibiotics while adhering to strict hygiene measures. DS bundles (including an informational video, a standard operating procedure, and ready-to-use kits for three cultures) can also improve outcomes and optimize BC diagnostics. Compared to new diagnostic methods, traditional bacterial culture has longer turnaround times and may lead to inappropriate antimicrobial use. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) is used for rapid microbial identification, characterization, and typing. However, it may detect Gram-negative bacteria better than Gram-positive bacteria. Routine application of this technology can also reduce mortality from bacteremia and further promote AMS. A randomized controlled trial evaluated the results of rmPCR detecting bacteria, fungi, and resistance genes directly from positive BCs, finding reduced use of broad-spectrum antibiotics without affecting mortality, length of stay, or costs. Another emerging diagnostic tool is next-generation sequencing (NGS), an easy-to-use, culture-free, PCR-based diagnostic method that appears to yield good results. Further research is needed to better understand the cost-effectiveness, impact on patient outcomes, and the role of novel diagnostic tools in managing BSI.5. Urinary Tract Infections The high pre-test probability of urinary tract infections (UTI) should be the primary driver for requesting urine culture (UC) orders. However, UCs are often performed in the absence of symptoms or with unclear symptoms, which may lead to overdiagnosis and inappropriate antibiotic treatment. Additionally, the yield of UCs may be compromised due to improper sampling, contamination, or misunderstanding. Clinicians should be aware that catheter-associated bacteriuria is common and usually indicates colonization rather than infection. Therefore, the IDSA’s 2019 clinical practice guidelines strongly recommend against screening UC in patients with indwelling catheters, and UCs should only be obtained from febrile patients at high risk for invasive infections (e.g., kidney transplant, recent urogenital surgery, neutropenic patients, or evidence of obstruction). To reduce unnecessary UC orders, many institutions have integrated computerized order entry and clinical decision support alerts into their electronic health systems (EHS), which automatically generate alerts whenever urine analysis, UC, or commonly used antibiotics for UTI are ordered. These strategies are optimized when combined with AMS educational support and guidance from infectious disease specialists. For example, including easily accessible guidelines in the EHS and requiring indications when requesting UCs has led to a 34% reduction in UC orders for catheterized patients. Similarly, when nurses reviewed UC orders for patients not meeting pre-specified criteria with intensivists, the neuro ICU reported a significant reduction in catheter-associated UTIs. UCs are only allowed when urine analysis (UA) meets pre-specified criteria (known as reflexive UC), which has been shown to significantly reduce unnecessary cultures. Pyuria on UA is the most important trigger for reflexive UC, with an NPV exceeding 90%. Other indicators used can include positive leukocyte esterase, positive nitrite, or >5-10 WBC/HPF. The presence of epithelial cells may indicate improper sample collection and contamination by skin flora, prompting physicians to reconsider UC treatment. Systemic biomarkers such as CRP and procalcitonin have limited or poor roles in UTI DS strategies. Other biomarkers, such as urinary myeloperoxidase, adenosine triphosphate, and urinary xanthine oxidase, have sensitivities and specificities too low to recommend. Novel diagnostic tools such as flow cytometry, MALDI-TOF-MS, and combinations of both have also been attempted but are limited by availability and cost. Conclusion Recent advances in AMS strategies aim to guide better patient care, improve clinical outcomes, while reducing unnecessary exposure to antimicrobials. DS is crucial for better implementation of stewardship activities. DS includes diagnostic strategies based on preset algorithms for testing and incorporates new diagnostic tools in patient evaluations. Despite some limitations and costs, these new diagnostic technologies have proven beneficial in the appropriate use of antibiotics across various clinical syndromes. Unfortunately, many low- and middle-income countries lack access to new molecular tests. Global collaboration among all stakeholders, including pharmaceutical companies, governments, and social organizations, is essential for better utilization of new technologies worldwide. Furthermore, close collaboration among infectious disease specialists, intensivists, and microbiologists in the hospital setting is necessary to optimize patient care in the ICU and provide evidence-based diagnostics and management.
