This is another one of those “practice-changing” types of articles, where the authors try to debunk some specific aggressive diagnostic or therapeutic modality that is over-utilized in a low-prevalence, high-risk population. This article, which you may have already seen, is regarding the need for a lumbar puncture in infants between 30 and 90 days.
They perform a retrospective review of nonconsecutive infants between 30 and 90 days of age who presented to the Emergency Department and received the “septic workup” – urinalysis/culture, blood culture, and lumbar puncture/CSF culture. They analyze a data set of 392 infants, the overwhelming majority of which are completely culture negative. 52 of them are culture positive on their urinalysis, 13 are culture positive in the blood, and 4 are CSF culture positive. The authors note that only one patient who had a positive urinalysis also had a positive CSF fluid culture – and that infant did not qualify as a low-risk infant by the Rochester criteria – so a well-appearing infant with a positive urinalysis need not undergo LP.
So, essentially, this study tells us only that meningitis is rare and that UTIs are common. The authors attempt to make the flawed logical argument against LP in their discussion by emphasizing the negative predictive value for meningitis in the setting of an abnormal UA is 98.2%. However, they erroneously discount the negative likelihood ratio of 0.87 (95% CI, 0.5–1.5). Therefore, statistically speaking, based on their results, repeating this study 100 times could lead to nearly half the study results showing a positive urinalysis favored concomitant meningitis.
Now, in a clinical sense, the authors are likely correct. An infant who looks well, meets the Rochester criteria, has an identified source for fever, and will be receiving antibiotics is at low risk for meningitis – by prevalence alone, not by anything this study shows – and is probable to have a good clinical outcome since they’re receiving antibiotics (in the event that same organism is resulting in a well-appearing, subclinical systemic and cerebrospinal bacteremia). The argument should not be that you can generate a zero-risk population with their combination of +UA and Rochester, but that the risk of bad outcome may be similar to the risk of harms associated with the lumbar puncture, false positives, and follow-on treatment/testing.