Discussion

We declared our 2020 case cluster a pseudo-outbreak based on a hypothesis that a culture media change improved detection of C. acnes in CSF at our institution and thus represented false-positive results. However, a stable C. acnes CSF culture rate after the media was changed permanently suggests the case cluster represented a true outbreak. Misclassification of the cluster led to a missed opportunity to identify the actual root cause of the outbreak and prevent future cases.

Pediatric CSF diversion device infections occur in up to 10% of patients depending on underlying risk factors and reinfection rates have been reported as high as 20%. ^{Reference McAlpine, Robinson, Barton, Balamohan, Davies and Skar8,Reference Kulkarni, Rabin, Lamberti-Pasculli and Drake9} The proportion of infections caused by C. acnes is poorly defined (1.4%–15% in some series). ^{Reference Everett, Eickhoff and Simon10,Reference Arnell, Cesarini, Lagerqvist-Widh, Wester and Sjölin11} The variability in diagnosing a C. acnes infection may be due to absence of typical symptoms and relatively slow growth of the organism in culture. ^{Reference Nisbet, Briggs, Ellis-Pegler, Thomas and Holland12,Reference Brook13} Although the 2017 IDSA guidelines recommend incubating CSF cultures for up to 10 days to promote growth opportunity of C. acnes, ambiguity remains whether this intervention would identify additional C. acnes infections or would increase the false-positivity rate of CSF cultures as most of our cultures resulted in <10 days. ^{Reference Tunkel, Hasbun, Bhimraj, Byers, Kaplan and Scheld2}

The clinical intervention for neurosurgical CSF diversion device infection involves extended hospitalization and immobilization for external CSF drainage, and prolonged antibiotic courses often requiring peripherally inserted central catheter (PICC) placement. ^{Reference Bignardi14,Reference Carter, Langley, Kuhle and Kirkland15} Revision of a neurosurgical shunt for any indication increases the risk of subsequent infection that rises with each additional revision. ^{Reference Simon, Butler, Whitlock, Browd, Holubkov and Kestle16} Inaction may not be benign as delayed diagnosis of neurosurgical shunt infections can have significant impact on patient outcomes, resulting in seizures, cognitive impairment, and death. ^{Reference Vinchon and Dhellemmes17,Reference Walters, Hoffman, Hendrick and Humphreys18} Thus, the decision to revise a neurosurgical CSF diversion device is far from trivial and introduces risks that are justifiable for high-pathogenicity organisms but may be less clear for low-pathogenicity organisms like C. acnes.

In this ambiguous context, we investigated the 2020 case cluster. The incidence of positive cultures was elevated, but the incidence of true infection was less clear. Clinical management of the seven patients was predominantly with observation. Only two of the seven cases were treated with antibiotics and one had surgical intervention although six had an infectious diseases consultation (data not shown). This suggested that clinical teams considered these positive cultures clinically insignificant and may have biased us to conclude we had a pseudo-outbreak.

Changes in anaerobic culture product use in the Microbiology Lab allowed us to reexamine our conclusion. We were surprised that C. acnes culture-positive rates returned to pre-outbreak levels and concluded that our 2020 cluster was indeed an outbreak. To understand impact on patients and utilization, as well as inform future responses to similar clusters, we reviewed charts of all patients with a C. acnes-positive CSF culture from 2016 to 2022. There was considerable variation in management, with the majority of cases receiving no interventions (Table 1).

Upon review, we concluded our surveillance system appropriately detected an increase in C. acnes cases, which is its goal. We faltered in our further analysis, however, when we considered the likelihood of case finding with a low-pathogenicity organism that is a ubiquitous colonizer. We did not pursue investigation tools such as molecular analysis of the isolates. However, should future clusters occur, we may include them earlier in the process in these high-risk populations even if the epidemiologic link is unclear to identify a source. ^{Reference Teoh, Ankrum, Meinzen-Derr, Weingartner, Goebel and Huang19} Thankfully, the outbreak self-resolved, but we cannot be sure the issue causing it was remediated.

We also concluded that guideline ambiguity around diagnostic criteria for infection likely contributed to our failure to recognize the outbreak. Contrary to IDSA guidance, holding our cultures for 10 days had little impact on our investigation or clinical care. Similarly, multiple consecutive positive cultures helpful for clinical decision-making would not influence the threshold to initiate outbreak investigation. Future guidelines may consider more stringent criteria to define true infection for C. acnes or other low-pathogenicity organisms. Guidance for oral suppressive regimens in patients who may not be good candidates for surgical intervention could support clinician decision-making and avoid surgical risks. Ultimately, greater guidance clarity would enable institutions to better discern device-associated outbreaks of indolent organisms like C. acnes earlier and intervene more effectively.

Our study is subject to limitations. Our delineation of pre-outbreak, outbreak, and post-outbreak time periods is arbitrary and may not represent optimal categorization. We had 12 months of post-outbreak data and are likely insufficiently powered to determine a difference. Since we observed a difference with the time periods chosen, shortening the outbreak period would likely generate higher infection rates and lead to a similar result. Additionally, we could not account for all potential changes in sample collection personnel or protocols in the pre-outbreak period. Nonetheless, our protocols have been standard in our institution for years, and many team members have had long tenures.