Discussion

To our knowledge, in the study, we have described the largest database worldwide with regard to EVD-associated meningitis/ventriculitis based on >220,000 EVD days from >150 ICUs. With an overall value of 3.96 EVD-associated meningitis/ventriculitis per 1,000 EVD days, this device-associated infection rate is higher than other device-associated infection rates. For example, in KISS, in the subgroup of neurosurgical ICUs, the VAP rate is 3.73 per 1,000 ventilator days, and the CAUTI rate and the CVC-BSI rate are much lower at 2.27 and 0.51 per 1,000 corresponding device days.⁹

Compared to the literature, our study found much lower overall EVD-associated infection rates. However, most of the studies providing these data are single-center studies, which may have a special interest in this topic due to their rather high infections rates.^{Reference Arabi, Memish and Balkhy10-Reference Hagel, Bruns, Pletz, Engel, Kalff and Ewald14} In addition, a systematic review including data from a total of 35 studies identified much higher overall EVD-associated meningitis/ventriculitis rate of 11.4 per 1,000 EVD days.^{Reference Ramanan, Lipman, Shorr and Shankar5} When interpreting our rather low EVD-associated infection rate, one should also consider that traditional surveillance based on manual review of patient charts is vulnerable to overlooking relevant diagnostic findings and may result in lower infection rates compared with the use of an automated surveillance method for meningitis/ventriculitis, as demonstrated by van Mourik et al.^{Reference van Mourik, Troelstra and Berkelbach15}

Most cultured organisms were gram positive (62.1%); staphylococci were the most common (47.2%), followed by Enterococcus spp (14,9%). In combination with gram-negative organisms (22.6%) and others (9.5%, including Corynebacterium, Bacillus, Micrococcus, Propionibacterium spp, Streptococcus spp, Haemophilus influenza, Morganella morganii, etc). These findings are analogous to those of other studies.^{Reference Hagel, Bruns, Pletz, Engel, Kalff and Ewald14,Reference Hoefnagel, Dammers, Ter Laak-Poort and Avezaat16,Reference Jamjoom, Joannides and Poon17} In the literature, an increase in the occurrence of gram-negative organisms in has been attributed to the use of antibiotic prophylaxis targeting gram-positive bacteria,^{Reference Martin, Zimmermann, Huynh and Polage18-Reference Alleyne, Hassan and Zabramski20} to the use of antimicrobially impregnated catheters,^{Reference Atkinson, Fikrey, Jones, Pringle and Patel21,Reference Bayston, Vera, Mills, Ashraf, Stevenson and Howdle22} and to prolonged hospitalization.^{Reference Martin, Zimmermann, Huynh and Polage18} Based on the surveillance method, we did not have access to individual patient data about these risk factors.

Also, an ICU type other than neurosurgical, surgical, interdisciplinary (>400 beds), or neurological was shown to be a protective factor in our multivariate analysis with regard to developing an EVD-associated meningitis/ventriculitis. Because the group of “other” ICUs mostly consists of interdisciplinary ICUs in small hospitals with <400 beds and, to a lesser extent, medical, cardiosurgical, burns, and pediatric ICUs, patients with an EVD might be not as severely ill and prone to an EVD-associated infection. However, we were unable to compare individual patient characteristics with regard to comorbidity and severity of illness and other risk factors like frequency of CSF sampling, location of EVD insertion (ie, ICU, operating room, or other), repeated insertions or duration of EVD. This shortcoming is a limitation underlying the surveillance method because the module ICU-KISS is unit or ward based. In this context, other data showed that an increased frequency of CSF sampling was a significant parameter associated with EVD-associated infection risk.^{Reference Hoefnagel, Dammers, Ter Laak-Poort and Avezaat16,Reference Jamjoom, Joannides and Poon17} The lack of data about EVD-associated meningitis or ventriculitis cases occurring after discharge from the ICU is an accompanying limitation, which might mask an actually higher overall EVD-associated infection rate.

High EVD utilization rate (above the 75th quantile) was a significant risk factor associated with increased infection rates. High EVD utilization rate could be a result of dealing with a rather special patient population being severely ill, necessitating EVD and thus over time having an increased risk of acquiring a device-associated infection. However, loose indications for EVD placement and removal could lead to unnecessarily high EVD utilization rates.

In our multivariate risk-factor analysis, we also investigated whether continuous participation in the surveillance system for a longer period of time (>3 years) was associated with a reduction of infection rates. However, this factor was not significant in our analysis and differed from our experience with VAP and CVC-BSI surveillance.^{Reference Schröder, Schwab and Behnke2,Reference Zuschneid, Schwab, Geffers, Rüden and Gastmeier23,Reference Zuschneid, Schwab, Geffers, Behnke, Rüden and Gastmeier24} Looking at all ICUs that either started surveillance of EVD-associated meningitis/ventriculitis in 2008 or later and continuously participated until 2016, the overall device-associated infection rate per 1,000 EVD days was ~4, without any downward trend over time. On the other hand, when investigating the surveillance effect on CAUTI, we detected an effect of surveillance not present before, including a large number of ICUs and a long observation period.^{Reference Gastmeier, Behnke, Schwab and Geffers25} Although the evidence has demonstrated that by applying appropriate infection control measures it is possible to decrease EVD-associated infections,^{Reference Leverstein-van Hall, Hopmans and van der Sprenkel12,Reference Scheithauer, Schulze-Steinen and Hollig13,Reference Chatzi, Karvouniaris and Makris26-Reference Camacho, Boszczowski and Freire28} other reasons for failure to reduce the infection rate in the previous studies might include a lower awareness of this entity and the lack of evidence-based guidelines on how to prevent EVD-associated infections compared with guidelines for the management of CVC, ventilators, and urinary catheters and the prevention of corresponding infections. In the light of the necessity of a comprehensive guidance for EVD management evidence-based guidance, documents have recently been produced on a national and local level.^{Reference Fried, Nathan and Rowe29-Reference Tunkel, Hasbun and Bhimraj32}

Meningitis and ventriculitis are relatively frequent and severe complications of care associated with the use of EVD compared with other device-associated infection rates. Therefore, and because of the high morbidity and mortality of these infections, surveillance systems should include EVD-associated meningitis/ventriculitis at least in neurosurgical ICUs and other ICUs with a substantial number of EVD days. As recommended for the prevention of other device-associated infections, attention should be paid to strictly assess the indications for EVDs aiming to minimalize device utilization. Analogously, strict adherence to hand hygiene and asepsis are crucial when handling the device and taking care of these patients.