Diagnostic Stewardship: It's all the rage!

Dan Morgan recently blogged here about diagnostic stewardship, referencing a JAMA viewpoint we published last year, and I’ll be presenting on the topic at the Remington Winter Course next month (join us!). So I wanted to draw attention to an excellent commentary just published in ICHE on diagnostic stewardship for healthcare-associated infections (HAIs), outlining opportunities and challenges. The key table is below. 

Also, in a great example of the need for diagnostic stewardship for HAIs, Clare Rock and colleagues just published this observational study in AJIC. They retrospectively reviewed 18 months of surveillance for hospital-onset C. difficile infection (HO-CDI) “LabID events” reported to NHSN. For those not acquainted with the NHSN LabID event metrics, they do not consider patient-level clinical variables—only lab results and admission/testing dates. Of the 490 HO-CDI cases that occurred during the study period, chart review determined that 206 (42%!) of them were not likely to represent “true” CDI. In about half of “untrue” cases there was no significant diarrhea (defined as >= 3 loose stools in 24 hours), in 41% the patient had received a laxative in the prior 48 hours, and in almost 10% of cases the symptom onset was prior to the “hospital-onset” criterion but testing was delayed. The graphs below demonstrate how improved test utilization could have changed their publicly-reported (and reimbursement-linked) HO-CDI rates. Of course the SIR data assumes that no other hospitals implemented similar diagnostic stewardship programs…

When I talk about diagnostic stewardship to my laboratory colleagues, they often seem a bit puzzled—“OK, please don’t test patients with low pre-test likelihood of disease—isn’t that just diagnostics 101?”. So why is the issue gaining more traction now? I think it’s due to advances in diagnostic technology and changes in health care delivery. Our tests are becoming more sensitive and expansive (e.g. “syndromic” panels that detect dozens of targets in one fell swoop), and at the same time clinicians are seeing more patients in shorter periods of time and have less time to think about the tests they order—leading to more reliance on technology and less reliance on the careful history and exam findings that are required to generate thoughtful assessments of pre-test disease likelihood. 

Also I used to trudge 5 miles through 12 inches of snow to get to grade school, and kids these days….

Finally, an apt comic from JCMs excellent new micro-comic series:

-->

The cartoon editorial, microbiology edition: An idea whose time has come!

I was excited to read the editorial in this month’s Journal of Clinical Microbiology (JCM), by Alex McAdam (JCM Editor in Chief), entitled “Prevalence and Predictive Values”. You can read it here too, because I’ve pasted it below:

Brilliant—a simple concept (diagnostics 101!) explained in a simple format. And as an associate editor of JCM, I can attest that this concept is frequently missed by submitting authors, not to mention practicing clinicians and hospital epidemiologists. 

This issue is also foundational to diagnostic stewardship, as it emphasizes the importance of limiting diagnostic testing to patients who have a reasonable pre-test likelihood of disease (pre-test likelihood being the individual-patient equivalent of population prevalence). 

It also explains why we’ll never “get to zero” for healthcare-associated infections (HAIs), even if all HAIs were preventable. Take the example of hospital-onset C. difficile infection (HO-CDI). The more successful your prevention program is at reducing whatever the “true” incidence of HO-CDI is, the lower will be the population prevalence—and the lower the positive predictive value (PPV) for the very sensitive CDI tests we now use. Positive tests will still occur, no doubt, and will be counted toward the HO-CDI rate—but they’ll be increasingly likely to be clinical false positives. 

Now I need to go start working on a good cartoon editorial about whether CAUTI exists….

Diagnostic Stewardship

The following is a guest post from Dr. Dan Morgan, GFOTB (Good Friend Of The Blog):

This week Preeti Malani, Dan Diekema and I wrote a viewpoint in JAMA discussing diagnostic stewardship, or “modifying the process of ordering, performing, and reporting diagnostic tests to improve the treatment of infections and other conditions.” In other words, guiding laboratory ordering to prevent contradictory results and reporting results in a fashion that makes treatment more appropriate. There really are two Criteria for Diagnostic Stewardship modifying laboratory testing:

1) Does it modify the process of ordering, performing and reporting tests? 

2) Does it improve the appropriateness of patient management? 

When I discussed diagnostic stewardship with my non-medical wife, she asked “you mean they don’t do that? Why would they do tests that contradict other results or provide second or third line antibiotic choices?” 

This is why I think diagnostic stewardship has so much potential. It is about making laboratory ordering more rational, which is hard to debate. Although medicine has existed with the idea that doctors knew best how to order and interpret results, we are now seeing they often don’t, as predicted by psychologists Danny Kahneman and Amos Tversky in the 1970s; “Intuitive judgments are liable to similar fallacies in more intricate and less transparent problems.” 

Doctors ordering and interpreting test results are like other people, often irrational. Diagnostic stewardship makes testing more logical to improve patient care. Ultimately this process shouldn’t be limited to urine cultures, blood cultures and C. difficile testing but applied to new molecular detection panels and non-ID tests, like cascading tests for anemia or limiting PSA testing in young and elderly men. And there has been interest in this idea from areas outside of ID. 

The fact that diagnostic stewardship reduces false-positive tests that contribute to publicly reported HAIs means there is likely a lot of incentive to support these processes. But we shouldn’t forget there are important patient benefits too, including avoiding unnecessary antibiotics, avoiding the distraction of misdiagnosis, and improving the ability of HAI rates to truly measure care.