Predictors and microbiology of ventriculostomy-associated infections: A retrospective study of 216 patients


 BACKGROUND AND PURPOSE: Ventriculostomy-associated infection (VAI) is a serious complication of external ventricular drainage (EVD). Infection rates and risk factors are variable in the literature due to heterogeneity of EVD procedures and management. This study aimed to investigate the incidence of VAI and its associated factors in two institutions with identical EVD procedural and management protocols.METHODS: The authors reviewed the medical records of patients older than 18 years who received EVD placements between January 2015 and December 2020 in two institutions under a single healthcare system. RESULTS: A total of 216 patients with 328 catheters was included for analysis. Twenty-three VAIs were identified, a 10.6% VAI rate. VAI was associated with a shorter duration of first EVD (9.7 days vs. 11.8 days, P = 0.018), a longer total EVD duration (28.0 days vs. 16.1 days, P < 0.001), a longer procedural time (72 minutes vs. 40 minutes, P < 0.001), catheterization at non-Kocher’s points (34.7% vs. 12.5%, P = 0.010), cerebrospinal fluid (CSF) leak (8.7% vs. 1.6%, P = 0.030), craniotomy (87.0% vs. 61.7%, P = 0.020), and other systemic infections (30.4% vs. 9.3%, P = 0.012). On multivariate analysis, the first EVD duration (OR = 0.095, P < 0.001), total EVD duration (OR 1.128, P < 0.001), EVD placement at non-Kocher’s point (OR 4.545, P = 0.012), and other systemic infections (OR = 5.117, P = 0.024) were associated independently with VAI. There was no statistical difference in VAI rate between patients with or without prophylactic EVD exchange at 14 days (6.8% vs. 12.6%, P = 0.247).CONCLUSIONS: Concomitant systemic infection and placement of an EVD catheter at a non-Kocher’s point were independently associated with VAI. Prophylactic EVD exchange at 14 days did not lower VAI rate.

the various and often contradicting results in the literature regarding the incidence of VAI and its risk factors.
The two institutions under the authors' healthcare system have a policy of prophylactic EVD catheter exchange every 10 to 14 days and routine CSF collection every three to ve days for infection surveillance. EVD placements performed between January 2015 and December 2020 were reviewed retrospectively and evaluated for prevalence of VAI and its risk factors.

Ethical Issues
This study was approved by the institutional review board of the participating institutions. The need to obtain informed consent from patients was waived because of the retrospective study design. The study was performed under the guidelines outlined by the Declaration of Helsinki, and the manuscript followed the STROBE checklist.

Study Design and Patient Population
The medical records of patients older than 18 years who received EVD placements between January 2015 and December 2020 in two institutions were reviewed retrospectively. A total of 412 patients received EVD placements during the study period. Patients who met the following conditions were excluded: incomplete medical records, pre-existing central nervous system infections prior to initial EVD placement, initial EVD placed in other hospitals, presence of more than two EVD catheters at any period of time, and death within 14 days from EVD placement. Finally, 216 patients with 328 EVD catheterizations were included for analysis. EVD Placement and Catheter Management All EVD placements included in this study were performed in the operating room. Prophylactic antibiotics, usually rst-generation cephalosporin unless the patient was already on other antibiotics, were administered in the operating room prior to skin incision. Skin preparation and surgical draping were performed in the usual manner. A standard EVD catheter without antibiotic-impregnation was used in all cases. When necessary, a navigation system was utilized to improve accuracy. The primary choice of EVD entry point was Kocher's point, and when necessary, EVDs were inserted at Paine's point, Frazier's point, or at the cisterna magna toward the third ventricle during craniotomy operations. Dressings were changed every two days at bedside, in an aseptic manner. A prophylactic catheter exchange was considered when an EVD catheter was expected to be maintained for longer than 14 days. When VAI was suspected, a catheter exchange was planned. CSF Collection and Detection of CSF Infection CSF samples were collected aseptically by surgeons or medical residents at the time of catheter insertion, routinely every three to ve days, and at the time of catheter removal. The number of CSF collections other than those performed at the time of insertion or removal of EVD catheters was recorded for each patient to re ect catheter manipulations. For practicality, VAI was diagnosed when positive CSF culture plus CSF protein > 40 g/dL and CSF glucose < 45 mg/dL were con rmed. Clinical Data Initial mental status at the time of EVD placement was dichotomized based on the Glasgow Coma Scale (GCS) as "good" if GCS was > 12 and "poor" if GCS was ≤ 12. Other systemic infections included sepsis with positive blood culture and relevant clinical signs, pneumonia, and urinary tract infection diagnosed with radiologic, laboratory, and clinical evidence. A CSF leak while the EVD catheter was in place was documented when there was de nite rhinorrhea or otorrhea following iatrogenic or traumatic skull base defect or CSF leak around the EVD insertion or lumbar drainage sites. Regarding the site of EVD catheterization, patients were dichotomized into having received EVD catheterizations only at Kocher's point and non-Kocher's point at least once. Timing of VAI was de ned as the interval between initial EVD catheterization and CSF collection from which a positive culture was observed. Data regarding the number of EVD catheter exchanges and fever above 38.0°C during EVD catheterization for each patient were collected. Functional outcomes were evaluated as 90-day modi ed Rankin Scale (mRS) being "favorable" if mRS ≤ 2, and "unfavorable" if mRS > 3. The primary endpoint was VAI detected from the CSF study, and secondary endpoint was 90-day ventriculo-peritoneal shunt (VPS) placement.

Statistical Analysis
Patients were divided into non-VAI and VAI groups, and data analysis was performed with IBM SPSS Statistics software Version 25.0 (IBM, Armonk, NY, USA). Univariate analysis was conducted using the Kruskal-Wallis test and Student t-tests for continuous variables and Fisher's exact test and Chi-square test for categorical variables. Multiple logistic regression was performed, and results were presented with odds ratio (OR) and 95% con dence interval (CI), with a two-sided probability value of 0.05 for signi cance.
Prophylactic antibiotic agents were used in all patients. The patient and procedure characteristics are summarized in Table 1. Factors associated with VAI VAIs occurred in 23/216 patients (10.6%). As presented in Table 2, age, sex, initial mental status, presence of IVH, underlying diagnosis, fever over 38.0°C, and number of CSF collections were not statistically different between VAI and non-VAI groups. On the other hand, a longer operation time was observed in the VAI group than in the non-VAI group (72 minutes vs. 40 minutes, P < 0.001), and the duration of the rst EVD catheter was signi cantly shorter in the VAI group (9.7 days vs. 11.8 days, P = 0.018), whereas the total duration of EVD catheterizations was signi cantly longer in the VAI group (28.0 days vs. 16.1 days, P < 0.001). The proportion of patients who received EVD placements via non-Kocher's point at least once was signi cantly higher in the VAI group (34.7% vs. 12.5%, P = 0.010). The proportions of patients with concomitant craniotomy surgeries during EVD placement, CSF leakage, and other systemic infection were signi cantly higher in the VAI group. Binary logistic regression was performed with variables with P-value < 0.10 and revealed that shorter duration of the rst EVD catheter, longer total duration of EVD catheterizations, EVD placement at a non-Kocher's point, and presence of other systemic infection were independently associated with VAI (Table  3).  (Table 4). For patients without prophylactic catheter exchanges, VAI rate was lower in those who maintained EVD ≥ 14 days than those who maintained EVD < 14 days, but the difference was not statistically signi cant (7.0% vs. 15.0%, P = 0.272).  12.6%, P = 0.247). The most commonly identi ed microorganism was CoNS, followed by gram-negative organisms such as Acinetobacter species and Enterococcus species.
VAI is a serious complication of the EVD procedure, resulting in prolonged hospital stay, unfavorable outcomes, and in-hospital mortality [10]. In general, three routes of CSF infection are suggested: 1) inoculation of skin ora during surgery, 2) manipulation during the CSF collection procedure, and 3) retrograde migration via the EVD tract [7]. In the current study, the numbers of CSF collections before VAI was identi ed between the VAI group and the non-VAI group were not statistically different (3.0 vs. 3.5, P = 0.337), consistent with a recent study by Sweid et al. who demonstrated that routine CSF collection and CSF collection when needed were not associated with VAI. [11] Our study suggests that CSF collection itself may not be a signi cant risk factor for VAI, provided that the closed CSF drainage system is not breached.
To reduce the risk of retrograde colonization, Mayhall et al. suggested prophylactic EVD exchanges in ve-day intervals [12]. However, various investigators questioned this recommendation. Holloway et al.
reported in their analysis of 584 patients with ventriculostomy that EVD catheters replaced prior to ve days did not lower infection rates compared to catheters replaced at longer intervals [13]. Later, a randomized controlled trial demonstrated that regular EVD changes at ve-day intervals failed to reduce VAI [14]. More recently, a study by Katzier et al. revealed that a group of patients who received elective EVD changes every ve days had a higher VAI rate than a group of patients who received EVD changes only when clinically necessary (32% vs. 8%). The authors suggested that a prophylactic EVD catheter change increased the incidence of VAI, and that inoculation of skin ora during the procedure was the primary contributing factor for VAI risk [7]. The researchers' analysis revealed that the prophylactic catheter exchanges in a 14-day interval did not lower VAI rates (6.8% vs. 12.6%, P = 0.247). The higher rate of VAI in patients without prophylactic EVD exchanges is because most VAIs occurred earlier than prophylactic catheter exchanges were considered, so that the patients with early VAIs belonged to the group of patients without prophylactic EVD exchange. Perhaps a more effective approach to control the risks of retrograde colonization or bio lm formation on the catheter surface than planning a catheter exchange at a certain interval is to implement EVD bundle approaches that consist of various infection prevention measures and use of silver-coated or antibiotic-impregnated EVD catheters [15][16][17][18]. Such actions have begun to be applied in the study institutions and were not re ected in this study.
In this study, the total duration of EVD catheterization was longer in patients with VAI than those without VAI (28.0 days vs. 16.1 days, P < 0.001). Whether the duration of EVD placement contributes to VAI is controversial. Several studies argued that prolonged EVD placement duration is a major risk factor for VAI [12,13,[19][20][21]. However, there are a few studies suggesting that duration of EVD placement is not associated with an increased risk for VAI. Lo et al. investigated 199 patients with 269 EVDs and recognized that the duration of drainage was not an independent risk factor for infection [8]. A similar nding was presented by Hagel et al., who demonstrated no signi cant difference in the duration between rst EVD placement and the occurrence of VAI and EVD removal in patients without VAI [22]. The authors noted that interpretation of previous studies that dealt with VAI was complicated because some studies used cumulative infection risks while other studies used daily infection risks. In the present study, the duration of rst EVD catheter was shorter in patients with VAI than those without (9.70 days vs. 11.81 days, P = 0.018), likely because infected catheters were replaced as soon as possible. Once VAI was noticed, a general practice was to maintain CSF drainage until three consecutive negative CSF culture results, performed in a two-to-three-day interval, were con rmed. Therefore, a prolonged EVD placement in this study was more likely to be the result of VAI, and not vice versa. Due to the multifactorial nature of VAI and the complexity of analysis of contributing factors, the in uence of catheterization duration on VAI is not clearly elucidated. However, as prolonged EVD catheter placement can interfere with early ambulation and discharge of patients, continuous efforts should be made to evaluate the necessity of maintaining EVD catheters in each patient and to remove the catheters as soon as possible.
On multivariate analysis, EVD catheterization via non-Kocher's point was an independent predictor of VAI.
32 patients received EVD catheter placements at non-Kocher's point, including Paine's point, Frazier's point, and at the cisterna magna to the third ventricle. Among them, 27 had EVD placements during craniotomy operations with long operation hours, exposing the surgical sites to a higher risk of microorganism inoculation. In addition, the authors suspect that EVD catheters inserted at entry points on the dependent position would be more di cult to maintain in an aseptic fashion, and this might contribute to the increased risk of VAI. Whether VAI risks differ by EVD entry point can be a topic of future studies.
CoNS species were the most commonly identi ed microorganisms (39.1%) in this study, consistent with several previous studies [6,8,22,23]. A notable nding is that Acinetobactor species (21.7%) and Enterococci species (17.4%) also were frequently identi ed. As in peripheral line-and central lineassociated bloodstream infections [24,25], the incidence and signi cance of gram-negative rods (GNR) in CSF infection have increased over the past few decades [7,8,21,26,27]. The increased proportions of Acinetobacter and Enterococcus could be in part related to the shift of microorganism populations in intensive care units toward gram-negative organisms [28,29]. The emergence and prevalence of multidrug resistant GNR are becoming a signi cant threat to the healthcare system. As prolonged use of prophylactic antibiotics targeting gram-positive organisms is considered one of the reasons for such a microbiologic shift [7], the bene ts and risks of prolonged use of prophylactic antibiotics in patients with EVD catheters should be reconsidered and re ected in clinical practice.
This study has several limitations. Its retrospective study design involved intrinsic limitations such as selection bias and incomplete data. As the protocols for placement and management of EVD catheters are heterogeneous and can differ by institution, the generalizability of our ndings might be limited. In addition, to diagnose VAI, patients' infectious signs and symptoms were not considered so it is impossible to completely exclude contaminations or colonization from true infections and to include culture negative VAIs, especially if prophylactic antibiotics were used in all cases. However, considering that it is often di cult or impossible to recognize signs and symptoms in patients with EVD catheters due to their deteriorated neurological status, the authors believe that their de nition of VAI was practical. Lastly, use of antibiotic-impregnated EVD catheters and implementation of the EVD care bundle, which have become popular aspects of EVD procedures for the control of VAI, would have allowed for more effective management of the risk of VAI and improved clinical outcomes.

Conclusion
This study demonstrated VAI in 23/216 patients (10.6%) and 23/328 catheters (7.0%). VAI was independently associated with a shorter duration of rst EVD catheter, a longer duration of total EVD, catheterization at a non-Kocher's point, and other systemic infections. Prophylactic EVD exchange at 14 days did not lower the risk of VAI.