Antimicrobial resistance (AMR) continues to escalate globally, reshaping clinical practice and threatening the foundations of modern medicine.1 In parallel, diagnostics in the clinical microbiology laboratory have undergone an unprecedented acceleration with the introduction of new techniques and automation. MALDI–TOF mass spectrometry, syndromic molecular panels, rapid disruptive phenotypic susceptibility tests, next-generation sequencing, and point-of-care devices are transforming how microbiologist identify pathogens and assess antimicrobial resistance.2,3 Moreover, artificial intelligence tools may likely improve efficiency and quality in clinical microbiology laboratories in the near future.4 These two forces, rising resistance and expanding diagnostic capacity, do not evolve independently. They collide at the bedside, shaping the decisions that determine patient outcomes. Antimicrobial stewardship programs (ASP) and diagnostic stewardship programs (DSP) were initially conceived as separate entities, each with its own culture, metrics, leadership, and operational logic, although despite the fact that the multidisciplinary nature of the teams involved meant that many of the professionals were represented in both programs.5
While ASP governs drugs use, DSP governs tests use. However, in clinical reality, antimicrobial prescription and diagnostic processes are not competing interventions, in fact they are two aspects of the same decision.6 Both programs were created at different times and with different metrics. ASP are well established, while DSP are just starting out. Although seemingly divergent in their origins, both must converge with the patient as the central focus of their mission in both its diagnostic and therapeutic aspects. These two programs show strong synergistic complementarity, providing comprehensive insight. The future of stewardship depends on reconceptualizing these programs as a unified diagnostic–therapeutic intelligence system rather than independent initiatives working in polite proximity.
Understanding the divergence requires revisiting the origins of each program. Antimicrobial stewardship emerged from epidemiological and clinical necessity: rising resistance, escalating drug consumption, ecological harm, and wide practice variability.7 Its priorities were clear: optimize antimicrobial use, reduce unnecessary exposure, shorten duration, and improve PK/PD alignment. Over time, antimicrobial stewardship acquired institutional legitimacy, multidisciplinary teams, and well-established metrics such as days of therapy or defined daily doses. Diagnostic stewardship, however, was born later and for different reasons. It developed as hospitals struggled to translate rapid technological advances into measurable clinical value. Diagnostic platforms multiplied faster than evidence supporting their impact. Diagnostic stewardship therefore emerged to ensure appropriate test ordering, selection of the test in the laboratory test, avoid low-value diagnostics, interpret results correctly, and link diagnostic information to actionable clinical pathways.8,9 Its vocabulary, turnaround time, analytical performance, test utilization, differed fundamentally from that of antimicrobial stewardship.10
The separation deepened because each program evolved within different professional cultures. ASP, led largely by infectious diseases specialists, preventive medicine specialist, microbiologists and pharmacists with particular skills on antimicrobials, was sometimes perceived as restrictive or punitive, a “police force” for antibiotics. DSP, led by microbiologists, was sometimes viewed as a gatekeeping entity controlling access to complex technologies. These caricatures, however simplistic, reflect structural realities: different budgets, reporting lines, workflows, and accountability systems. Most importantly, ASP and DSP traditionally used metrics that captured their internal performance rather than their shared clinical objectives. Reducing antibiotic use is not inherently beneficial unless tied to improved outcomes. Lowering diagnostic volume is not a success if it delays or impairs clinical decisions. As long as metrics remained compartmentalized, the programs remained operationally divergent.
Yet at the bedside, there is no such divergence. A diagnostic test has value only if it improves treatment decisions or impact hospital management decisions. An antibiotic has value only when supported by timely and accurate diagnostic information. The clinical space where patient care occurs dissolves the artificial borders between ASP and DSP. Convergence therefore becomes not an option but an inevitability. Both programs aim to reduce uncertainty, accelerate effective therapy, avoid unnecessary interventions, and minimize ecological harm.5 When embedded within shared clinical pathways such as sepsis, pneumonia, febrile neutropenia, or bloodstream infections, antimicrobial stewardship and diagnostic stewardship programs naturally operate in synchrony.11 Multiple studies have consistently demonstrated that delays in the initiation of appropriate antimicrobial therapy are associated with increased mortality in the severe infections described above.12,13 Prompt microbiological diagnosis is therefore critical to guide timely and targeted treatment.14,15 The challenge is to formalize this synchrony through governance, workflow integration, and shared metrics.
Metrics for integrationNew conceptual tools are essential for the integration of ASP and DSP. Among these, the introduction of “return on investment per diagnostic hour” (ROI/h) represents a fundamental shift. Traditional ROI calculations fail to capture the temporal value of diagnostics. ROI/h reframes the discussion: how much clinical or economic benefit is produced for each hour of diagnostic acceleration?16 As an example, a rapid syndromic molecular panel that saves forty hours of clinical uncertainty, facilitates earlier de-escalation, and reduces ICU days may show a dramatically higher ROI/h than a cheaper, slower test. Conversely, a test that shortens time-to-result by only two hours, even if analytically impressive, may have minimal clinical impact. This metric replaces the notion of “cost of a test” with the more clinically meaningful concept of “value per hour gained.” It is a language that both antimicrobial stewardship and diagnostic stewardship programs can speak, and one that hospital leadership can immediately interpret.
Another transformative metric is the diagnosis-to-treatment delta (ΔD→T), which measures the time between the availability of an actionable diagnostic result and the corresponding therapeutic intervention.17 ΔD→T shifts attention away from laboratory performance alone and toward the entire clinical workflow. A positive blood culture with rapid molecular identification delivered in one hour is clinically meaningless if the antimicrobial change occurs twelve hours later because no one reviewed the result or because logistical delays impeded treatment. High ΔD→T values expose hidden bottlenecks—communication failures, pharmacy delays, incomplete coverage models—that neither ASP nor DSP can resolve independently. Minimizing ΔD→T becomes the shared key performance indicator (KPI) of integrated stewardship, because it captures whether data truly become action.
Diagnostic time-outs provide a complementary framework.18 Modeled after antibiotic time-outs at 24–48h, they require clinicians to review what tests have been performed, what results are available, which ones are unnecessary, and how new information modifies diagnostic probability. They prevent cascades of low-yield tests and reinforce the principle that diagnostics must refine, not inflate, clinical reasoning. Adaptive testing strategies expand this idea further: instead of static catalogs, dynamic test-ordering platforms adjust options based on clinical context, preliminary findings, seasonality, epidemiology, and patient-specific factors. These systems reduce variability in ordering behavior, elevate pretest probability, and prepare data streams for artificial intelligence integration.
Future directions for the integration of antimicrobial stewardship and diagnostic stewardship programsOne of the most innovative concepts is “bidirectional authorization,” a symmetry of stewardship power. Microbiology laboratories can legitimately decline third or fourth unnecessary repeat tests, while infectious diseases teams of pharmacists can legitimately deny unjustified antibiotic prolongations. This reciprocal authority introduces constructive friction that improves clinical quality. It affirms that stewardship is not hierarchical but collaborative, with each discipline empowered to prevent low-value care in its domain.
The future, however, extends far beyond improved collaboration. The next decade will witness the emergence of fully integrated diagnostic–therapeutic stewardship systems aligned with artificial intelligence (AI) tools.19,20 This model fuses ASP and DSP into a single operational intelligence layer supported by artificial intelligence, decision-support algorithms, dynamic epidemiological models, and real-time clinical learning. Diagnostic–therapeutic stewardship combined with AI requires a new professional culture. Infectious diseases physicians must become fluent in diagnostic interpretation, while clinical microbiologists must adopt a more clinical, outcome-oriented perspective. Cross-training will be essential, as will a shift from narratives of ownership (“who leads”) to narratives of purpose (“which decision improves outcomes”).
AI will become central to diagnostic–therapeutic stewardship program. Integrative models capable of combining microbiology results, biomarkers, electronic health records, antimicrobial pharmacokinetics, colonization history, and local epidemiology will be able to predict likely pathogens and resistance patterns, even before culture results are available.21,22 They will recommend personalized diagnostic strategies, predict therapeutic efficacy, and continuously adapt based on local feedback loops. Theoretical probabilistic antibiograms will replace static ones, generating patient-specific susceptibility predictions hours before traditional testing.23 Digital infection twins, dynamic computational models integrating microbiome data, pharmacogenetics, PK/PD predictions, and epidemiological context, will guide individualized treatment and diagnostic pathways.24 Hospitals will increasingly function as adaptive ecosystems in which each diagnostic–therapeutic cycle refines local predictive models. In this environment, stewardship becomes not a committee but an operating system.
ConclusionsASP and DSP were born from different needs, shaped by different cultures, and measured by different metrics. Nevertheless, their future is inseparable. The question is no longer whether they diverge or converge. They must converge because the patient's diagnostic–therapeutic journey is indivisible. The objective of stewardships programs is not to reduce antibiotic use or reduce test volume; it is to optimize clinical decisions with maximal efficacy and minimal ecological damage. Diagnostic–therapeutic stewardship combined with AI represents the next stage: an integrated, intelligent, self-learning framework that synchronizes diagnosis, treatment, and knowledge (Fig. 1). In this model, hospitals stop treating stewardship as a program and start recognizing it as the core architecture of safe, modern clinical microbiology and infectious disease practices.
Conceptual framework of diagnostic–therapeutic stewardship-AI, illustrating the integration of rapid diagnosis, optimal therapeutic decision-making, and real-time learning within a unified diagnostic–therapeutic intelligence model. The figure was created using artificial intelligence (ChatGPT 5.1 Image Generation) according to the authors’ instructions.
During the preparation of this work the author(s) used ChatGPT 5.1 Image Generation to create the figure according to the authors’ instructions. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article.
Conflict of interestsThe authors declare having no conflict of interests.
