Discussion

Limiting CSF tests available to physicians did not significantly affect costs or clinical outcomes in patients with suspected meningitis. Despite the reduced unapproved testing rate, the cost and number of performed tests were not statistically different due to increased utilization of the FA/ME panel in the post-intervention period. Since the FA/ME panel is done in-house, our intervention allowed physicians to receive CSF testing results faster without increasing total laboratory costs.

While our intervention did not show laboratory cost savings with increased FA/ME panel use, a meta-analysis showed potential cost savings with FA/ME use through metrics such as length of stay. ^{Reference Hueth, Thompson-Leduc and Totev6} Conversely, another study showed that increased FA/ME panel use led to increased direct testing costs but lowered antimicrobial costs, creating no net change in the total cost of care. ^{Reference Soucek, Dumkow, VanLangen and Jameson7} Our intervention differed by increasing FA/ME panel utilization without increasing laboratory costs by requiring ID consult for other testing. Still, there were no associated changes in antibiotic prescriptions or LOS. The increase in FA/ME utilization mediated the lack of cost reduction at a cost of 854 USD at our institution, which was higher than previous literature at 214.44 USD (United States hospital) and 220 CAD (Canadian hospital). ^{Reference Soucek, Dumkow, VanLangen and Jameson7,Reference DiDiodato and Bradbury8} Future improvements to our intervention include panel limitations based on CSF white blood cell counts, which may safely decrease panel utilization without affecting clinical outcomes. ^{Reference McCreery, Nielsen, Clarey, Murphy and Van Schooneveld9,Reference Broadhurst, Dujari, Budvytiene, Pinsky, Gold and Banaei10}

Limitations to our study include being at a single center and our inability to account for seasonal variation over ten months. We were unable to assess whether the increase in FA/ME panel usage limiting cost savings was due to our intervention or increased clinician familiarity with the test. Clinicians were provided a list of approved assays including the FA/ME panel through an email detailing the intervention protocol, which could have prompted physicians to preferentially order all seven of the approved tests. Our institution’s cost of the FA/ME panel was higher than in previous literature, which could prevent cost reduction from our intervention. ^{Reference Hueth, Thompson-Leduc and Totev6,Reference Soucek, Dumkow, VanLangen and Jameson7}

It is important to note that our intervention could create additional ID consults as required for CSF testing outside of the pre-approved diagnostics. Although we are unable to measure total consult numbers at our institution, this effect may be tempered by cases requiring nonstandard ID testing already necessitating ID physician consultation. Limited evidence also shows that the FA/ME panel could have lower sensitivities than individualized viral polymerase chain reaction (PCR) tests, which prevents fully removing all viral PCR tests that overlap with FA/ME panel coverage. ^{Reference Broadhurst, Dujari, Budvytiene, Pinsky, Gold and Banaei10}

Our study’s strengths include its large sample size compared to previous studies. ^{Reference Hueth, Thompson-Leduc and Totev6,Reference Soucek, Dumkow, VanLangen and Jameson7} We also employed multiple regression models to control for confounding variables. Finally, our quasi-experimental design with defined intervention periods strengthens our study by allowing us to assess the intervention’s impact in a real-world setting.