Diagnostic Stewardship Interventions That Make a Difference

Aug. 3, 2021

"Medicine is a science of uncertainty and an art of probability."

Daniel Morgan, M.D. opened his World Microbe Forum talk on diagnostic stewardship with this quote from Sir William Osler that rings true with anyone in the field of clinical diagnostics. Setting the stage for all talks that followed, Morgan then provided a sweeping overview of the concept of diagnostic stewardship from the perspective of a clinician and hospital epidemiologist—and with the uncertainty of medicine and art of probability woven throughout.

Morgan highlighted a problematic misconception of both the general public and clinical providers: that the result of a diagnostic test undoubtedly determines the presence or absence of disease. A positive result makes a disease or condition more likely, but not certain, while a negative result makes a disease or condition less likely. Diagnosing disease involves the delicate balancing of probabilities. Before obtaining any testing, the pretest probability of having a disease must be considered. The test result will change this probability, but tests and sample collection are imperfect, so the result will rarely make the probability 0% or 100%. For example, an infectious disease test can be positive without actual infection, due to residual RNA or DNA in the specimen, contamination or assay complications. Alternatively, an infectious disease test can be negative when a person is truly infected. Common reasons for this include inadequate sampling and transport practices or the timing of collection.

Image of Fagan's Nomogram showing pretest probabilities and posttest results.
Fagan's Nomogram showing how the pretest probability and likelihood ratio of a positive diagnostic test lead to a posttest probability of 70% (black), while the pretest probability and likelihood ratio of a negative diagnostic test lead to a posttest probability of 0.50% (red).
Source: Modified from wikimedia.org.

The rapid advancement of diagnostic technology has come with a set of unique challenges. While ordering tests in the face of low pretest probability (i.e., the patient is unlikely to actually have the disease being tested for) is not a new problem, it is exaggerated by the high sensitivity of new molecular technologies, which also typically identify multiple targets at once. When these tests are used in the setting of low pretest probability, the results are challenging to interpret and likely lead to overtreatment. Specifically, the identification of a bacterial target that is likely not the cause of a patient's symptoms can lead to unnecessary antibiotic prescribing. To better navigate these challenges, diagnostic stewardship aims to improve the ordering, collection, processing and reporting of diagnostic tests to improve patient management and reduce the excessive use of antibiotics. Ultimately, diagnostic stewardship aims to ensure that the correct test is ordered on the right patient at the right time, and that it prompts the right action.

Most tests are likely over-ordered. In fact, the history and physical a clinician takes during an exam are responsible for 73%-94% of the information needed to give a diagnosis, with diagnostic testing making up far less than that. To complicate matters, clinicians tend to overestimate pretest probability before they ever order a test. Morgan referenced a study he led in 2021 that assessed clinicians' abilities to determine the probability of disease before and after testing. This survey study demonstrated that in common primary care practice scenarios, clinicians overestimated the probability of disease 2-10 times compared to available scientific literature and rarely adjusted this overestimation after the test result came back. These findings call into question the usefulness of the diagnostic test and the provider's clinical decision-making skills. Among many others in the literature, these findings support the need for diagnostic stewardship at all stages of the diagnostic process.

Donna Wolk, Ph.D, Director of Clinical Microbiology at Geisinger Laboratories, described a concept developed at Geisinger called the "true north" of diagnostic stewardship, designed  to help guide what diagnostic stewardship ideally entails in order to optimize care. This Venn diagram, in which "true north" is the center, comprises components essential to excellent care delivery:

  • Diagnostic intelligence: diagnostic stewardship, decision support and risk stratification.
  • Diagnostic readiness: diagnostic subject matter expertise and value-based care.
  • Diagnostic interventions: diagnostic speed and accessibility.
  • Diagnostic impact: antimicrobial stewardship, radiologic stewardship, bed availability, general resource recovery, safety, quality and financial stewardship.

Additionally, Wolk highlighted vital concepts to be considered when implementing a diagnostic stewardship program or implementing a rapid diagnostic test:

  1. Support: The new test or program should support The Joint Commission requirements for inpatient and outpatient antimicrobial stewardship programs.
  2. Actionable: Test results should be actionable in both the inpatient and outpatient settings where they will drive therapy, further testing, discharges, etc.
  3. Reduce: Testing should support shorter emergency department wait times and/or reduced emergency department visits.
  4. Population health: From the population health perspective, onsite testing with therapy should support reduced infection transmission.
  5. Precision medicine: From the precision medicine perspective, testing and therapy should follow algorithms to the patient with risk stratification.

Microbiology-Led Stewardship Research

Upon reviewing their quality metrics, Wolk and her team at Geisinger noticed they had a respiratory virus season problem. Patients seeking care for influenza-like illness in the outpatient setting were not getting test results within the designated turnaround time window. Consequently, a lack of timely and accurate respiratory test results led to antimicrobial overuse and additional costs to the healthcare system.

To study the impact of this problem, the Geisinger team launched the FluWorks© project, which compared the downstream outcomes at sites where onsite RT-PCR testing was implemented versus those that couriered specimens to a central laboratory. Overall, the impact of implementing rapid respiratory diagnostics at outpatient clinics was impressive. The median collect-to-result time decreased by 82% in the point of care RT-PCR group. In addition, clinicians were less likely to prescribe an antiviral medication before and after testing, saving 674 antiviral doses, equating to nearly $75,000. Almost $170,000 was saved between antivirals, radiologic imaging and other costs when all was said and done.

Among several others performed at Geisinger, this microbiology-led study serves as a prime example of how clinical microbiologists can perform hypothesis-driven or pragmatic quality studies with a significant impact. The microbiology laboratory may be the first to know when changes in infection prevalence occur or quality patterns (in the way tests are used and sent to the laboratory) are observed, and their subject matter expertise is imperative for infection prevention initiatives and interdisciplinary teams.

Diagnostic Stewardship and Broad-Range PCR

Bacterial broad-range PCR specifically targets conserved regions of bacterial DNA to identify pathogens present in clinical specimens. This technology was developed and used for taxonomic purposes in the 1980s, but was not in routine clinical use until 2015. Initial challenges related to testing implementation, such as lack of detail in reference libraries, cost and automation of the 16S PCR process, have since been resolved. The focus now is on the appropriate utilization of such testing and how it impacts patients. Rebekah Dumm, M.D. and her team at the University of Pennsylvania looked at 3 questions: First, what is the University of Pennsylvania's experience with this testing? Second, can the ideal specimen for this test be identified, optimizing the utility of the test? Finally, how do broad-range PCR results impact clinical management?

A faint, fastidious organism (Granulicatella) seen in the Gram stain from brain tissue.
A faint, fastidious organism (Granulicatella) seen in the Gram stain from brain tissue. Sterile specimens like brain with organisms seen in the Gram stain are significantly more likely to be positive during broad-range PCR testing. This may be particularly helpful if the organism does not grow well in culture.
Source: Andrea Prinzi

The researchers developed a scoring system to determine the predictive value of previously available patient data, which included information about pathology, culture and Gram stain results, compared to broad-range PCR results. This score, called the Visualized Inflammation/Organism Score System (VIO), ranks results from 0-4 based on the presence of inflammation or organisms reported from pathology or Gram stain. Interestingly, the VIO score corresponded with broad-range PCR results (i.e., in specimens with a low VIO score, broad-range positivity was limited). For specimens with organisms reported on Gram stain (which are likely to have a higher VIO score), the odds of the broad-range PCR being positive were 80 times greater.

Overall, broad-range PCR positivity was highest in heart valve, spinal fluid and prosthetic joint specimens. Specimens with a low VIO score had a low likelihood of PCR positivity, and negative PCR results from this group were unlikely to alter the clinical management plan in any way. Unsurprisingly, specimens with a high VIO score were associated with a high likelihood of positive PCR results. The results were often used to confirm an existing infectious diagnosis. The findings from this study can guide diagnostic stewardship efforts toward appropriate ordering of the test, including which specimen types may yield the most helpful information and how to incorporate the findings from readily available tests, such as Gram stain, in PCR result interpretation.

High-Impact, Low-Effort Interventions

Given that diagnostic stewardship can occur at various stages in a diagnostic test life cycle, it is possible to make minor adjustments that pack a powerful punch. Austin Ing, Pharm.D. and his team at Vanderbilt University Medical Center (VUMC) demonstrated this in a recent intervention they performed in the microbiology laboratory.

Coagulase-negative staphylococci (CoNS) growing on blood agar.
Coagulase-negative staphylococci (CoNS) growing on blood agar after being sub-cultured from a positive blood culture bottle.
Source: Andrea Prinzi

After reviewing data, the research team determined that 34% of positive blood cultures at VUMC grow coagulase-negative staphylococci (CoNS), and less than 25% of those with CoNS represent true infection. Treatment of CoNS contaminants has been associated with inappropriate antimicrobial use, prolonged length of stay and increased costs. The research team wanted to assess the impact of not reporting CoNS susceptibilities on vancomycin use. Historically, VUMC performed susceptibility testing on all isolates and reported these results in the medical record. The researchers looked retrospectively at vancomycin utilization in patients who had susceptibilities reported in their medical record, compared to patients who had interpretive guidance for CoNS blood culture contamination reported rather than susceptibility testing.

Ultimately, lack of CoNS susceptibility reporting did not significantly decrease overall use of vancomycin. However, there was a significant decrease in vancomycin trough testing, a method used to measure the amount of vancomycin circulating in the bloodstream that helps determine if the patient is receiving a therapeutic dose, and which is associated with additional blood draws and cost to the patient. While the number of patients started on vancomycin did not change, there were fewer doses (lesser duration) given in intensive care unit patients. These findings suggest that by making a relatively simple procedure change in the microbiology laboratory, some risks associated with excessive antibiotic use, such as drug reactions, may be reduced and substantial cost savings can be incurred.

Clinical Labs Are Natural Leaders for Diagnostic Stewardship

Although thousands of microbiology tests are ordered and interpreted daily and may seem routine, the process is complex and regularly contributes to diagnostic error. Diagnostic stewardship is a growing field where the clinical microbiology laboratory can be an exceptional leader. As demonstrated by the fantastic research presented at World Microbe Forum, clinical laboratories can be essential members of multidisciplinary stewardship teams. They can lead hypothesis-driven studies that demonstrate the impact of diagnostic stewardship interventions and optimize testing in ways that improve patient care and reduce harm.

Author: Andrea Prinzi, Ph.D., MPH, SM(ASCP)

Andrea Prinzi, Ph.D., MPH, SM(ASCP)
Andrea Prinzi, Ph.D., MPH, SM(ASCP) is a field medical director of U.S. medical affairs and works to bridge the gap between clinical diagnostics and clinical practice.