relative risks of ventriculostomy infection and morbidity

6
Acta Neurochir (Wien) (1994) 127:79-84 =Acta N&rochirurgica Springer-Verlag 1994 Printed in Austria Relative Risks of Ventriculostomy Infection and Morbidity C. G. Paramore 1 and D. A. Turner 1 3 Departments of I Surgery (Neurosurgery), 2 Neurobiology, and 3 Duke University Medical Center, Research and Surgery Services, Durham VAMC, Durham, NC, U.S.A. Summary Ventricular catheter placement is a common procedure for the management of increased intracranial pressure. Hypotheses regard- ing the etiology of infection of catheters center on two alternative assumptions: 1) contamination leading to infection occurs at the time of catheter insertion, implying that catheter duration has min- imal effect on infection risk; and 2) infection of catheters derives from catheter contamination after insertion, suggesting that duration of catheter use may significantly affect infection risk. We have studied the relative complication rate of ventricular catheter insertions using a retrospective approach (n = 161 patients and 253 catheter insertion procedures). The overall infection rate was 4.1%, but the daily infection hazard increased exponentially with time, to a maximum daily rate of 10.3% by day 6 of catheter insertion. This increasing risk appears most consistent with the sec- ond hypothesis. The risk of non-infectious complications was 5.6%, including hemorrhagic occurrences and misplacement severe enough to require a new catheter insertion. The daily hazard of infection approximately equalled the non-infectious risk of routine catheter replacement by day 5. Additional prospective data on the daily risk of CSF infection and the appropriateness of antibiotic prophylaxis either at the time of ventricular catheter insertion or continued through the catheter's presence may be required to both definitively identify which hy- pothesis of infection risk is correct and whether antibiotics can sig- nificantly ameliorate this risk. Keywords: CSF Drainage; ICP recording; trauma; ventriculos- tomy complications. Introduction Ventricular catheters are a valuable adjunct for the measurement and management of raised intracranial pressure (ICP) and continue to be the standard to which other devices are compared 9' l l, 12, ~4,15. However, there exists no consensus in the literature regarding the op- timum time, if any, for routinely replacing a functioning ventricular catheter in a patient who requires continued monitoring and/or CSF drainage. The'risk of infection has commonly been used as the basis for conclusions regarding the need to replace ventricular catheters. The infection rate for ventriculostomy is reported to be between 0 and 27% in a large number of retrospective and prospective clinical series 1' 2, 5, 8-14, 16-21, Several reports suggest that the risk of infection is independent of the length of catheter duration, implying that a work- ing catheter should be left in place as long as it is needed and remains uninfected4' 13, !6-20. Conversely, other au- thors have suggested that the risk of infection increases significantly with the time after insertion 7' m, 12,2i. Par- ticularly, a prospective study by Mayhall et all ~ indi- cated that catheters should be routinely replaced after five days to avoid a significant risk of infection< 1~, 19 However, these studies have focused on infectious risks alone and have not considered the risks of catheter replacement, particularly in patients with shift or col- lapsed ventricles where replacement may be difficult. In addition, ventriculostomy systems are usually treated as closed and are tunneled. Thus, the risks of both infection and replacement may be quite different than those associated with systemic lines used for both frequent access and monitoring, such as arterial blood pressure or CVP catheters. Data concerning other complicating factors in ven- tricular catheter placement are scant. Hemorrhage ap- pears uncommonly following catheter insertion (< 2% 12, ~9) and the accuracy of catheter placement is usually not specified. It would be desirable to incor- porate both these mitigating factors and the risk of infection into a balanced determination of relative risks to ascertain when to routinely replace a functioning ventricular catheter. We have performed a retrospective analysis of all patients undergoing ventricular catheter

Upload: c-g-paramore

Post on 10-Jul-2016

214 views

Category:

Documents


2 download

TRANSCRIPT

Acta Neurochir (Wien) (1994) 127:79-84 =Acta N&rochirurgica

Springer-Verlag 1994 Printed in Austria

Relative Risks of Ventriculostomy Infection and Morbidity

C. G. Paramore 1 and D. A. Turner 1 3

Departments of I Surgery (Neurosurgery), 2 Neurobiology, and 3 Duke University Medical Center, Research and Surgery Services, Durham VAMC, Durham, NC, U.S.A.

Summary

Ventricular catheter placement is a common procedure for the management of increased intracranial pressure. Hypotheses regard- ing the etiology of infection of catheters center on two alternative assumptions: 1) contamination leading to infection occurs at the time of catheter insertion, implying that catheter duration has min- imal effect on infection risk; and 2) infection of catheters derives from catheter contamination after insertion, suggesting that duration of catheter use may significantly affect infection risk.

We have studied the relative complication rate of ventricular catheter insertions using a retrospective approach (n = 161 patients and 253 catheter insertion procedures). The overall infection rate was 4.1%, but the daily infection hazard increased exponentially with time, to a maximum daily rate of 10.3% by day 6 of catheter insertion. This increasing risk appears most consistent with the sec- ond hypothesis. The risk of non-infectious complications was 5.6%, including hemorrhagic occurrences and misplacement severe enough to require a new catheter insertion. The daily hazard of infection approximately equalled the non-infectious risk of routine catheter replacement by day 5.

Additional prospective data on the daily risk of CSF infection and the appropriateness of antibiotic prophylaxis either at the time of ventricular catheter insertion or continued through the catheter's presence may be required to both definitively identify which hy- pothesis of infection risk is correct and whether antibiotics can sig- nificantly ameliorate this risk.

Keywords: CSF Drainage; ICP recording; trauma; ventriculos- tomy complications.

Introduction

Ventricular catheters are a valuable adjunct for the measurement and management of raised intracranial pressure (ICP) and continue to be the standard to which other devices are compared 9' l l, 12, ~4, 15. However, there exists no consensus in the literature regarding the op- timum time, if any, for routinely replacing a functioning ventricular catheter in a patient who requires continued monitoring and/or CSF drainage. The'risk of infection

has commonly been used as the basis for conclusions regarding the need to replace ventricular catheters. The infection rate for ventriculostomy is reported to be between 0 and 27% in a large number of retrospective and prospective clinical series 1' 2, 5, 8-14, 16-21, Several

reports suggest that the risk of infection is independent of the length of catheter duration, implying that a work- ing catheter should be left in place as long as it is needed and remains uninfected 4' 13, !6-20. Conversely, other au- thors have suggested that the risk of infection increases significantly with the time after insertion 7' m, 12, 2i. Par- ticularly, a prospective study by Mayhall et a l l ~ indi- cated that catheters should be routinely replaced after five days to avoid a significant risk of infection < 1~, 19 However, these studies have focused on infectious risks alone and have not considered the risks of catheter replacement, particularly in patients with shift or col- lapsed ventricles where replacement may be difficult. In addition, ventriculostomy systems are usually treated as closed and are tunneled. Thus, the risks of both infection and replacement may be quite different than those associated with systemic lines used for both frequent access and monitoring, such as arterial blood pressure or CVP catheters.

Data concerning other complicating factors in ven- tricular catheter placement are scant. Hemorrhage ap- pears uncommonly following catheter insertion (< 2% 12, ~9) and the accuracy of catheter placement is usually n o t specified. It would be desirable to incor- porate both these mitigating factors and the risk of infection into a balanced determination of relative risks to ascertain when to routinely replace a functioning ventricular catheter. We have performed a retrospective analysis of all patients undergoing ventricular catheter

80

placements over a three-year period at Duke University Hospital, to evaluate the frequency of both infectious and non-infectious complications and to determine if the routine replacement of ventricular catheters was warranted.

Clinical Methods

The charts of 161 patients on the Neurosurgery service at Duke University Hospital who received 253 ventricular catheter placements (between January 1, 1989 and December 31, 1991) were reviewed for the following parameters: admission diagnosis, location of cath- eter placement (operating room, neurological ICU, etc.), number of days the ventriculostomy was in place, complications, CSF white blood count and CSF culture results. Ventricular catheters were placed using the tunnel technique (Friedman et al. 5) in the intensive care units, the emergency room or the operating theater. Prophylactic antibiotics, either Nafcillin or Vancomycin, were used in all cases for the catheter duration and were given in addition to any other antibiotics the patient was receiving for co-incident infections. This total group (253 catheter placements) was analyzed for non-infec- tious complications, defined as either the occurrence of an obvious decline in neurological status immediately following catheter place- ment or the presence of hemorrhage or misplacement of the catheter on a subsequent computed tomography (CT) scan. Many of the ventriculostomy catheters were routinely changed at day 5 following insertion, limiting the study beyond this point.

A subgroup of 135 patients (who underwent 220 separate catheter placements) was also analyzed for infectious complications, exclud- ing patients who either were treated for a CSF shunt infection or who expired within 24 hours of catheter placement. Infection was defined as a positve CSF culture from either the ventriculostomy catheter or lumbar subarachnoid space, verified by growth on agar plates. Cerebrospinal fluid was routinely drawn from ventriculos- tomies only at the time of catheter removal, or when an infection was clinically suspected. Thus, the primary source of culture data was obtained from the routine withdrawal cultures from all catheters. The date of the first positive culture was recorded as the date infection occurred. The infection data were statistically analyzed using a Ka- plan-Meier analysis (with infection as the primary end point), which provided a daily hazard for infection. The day of the first positive culture was used as the end point for infected catheters. Regression analysis was used to determine the function best fitting the daily

Table 1. Admission Diagnoses

Diagnosis Number of patients

Closed head injury 40 Subarachnoid hemorrhage 37 Intracerebral hemorrhage 28 Shunt infection 13 Hydrocephalus 12 Epidural or subdural hematoma 7 Intraventricular hemorrhage 6 Other 18

Total 161

C. G. Paramore and D. A. Turner: Relative Ventriculostomy Risks Table2. Pathogens in Infected Ventricular Catheters

Pathogen Number of cases

Staphylococcus coagulase ( - ) 5 Staphylococcus coagulase (+ ) 1 Klebsiella pneumonia 2 Enterobacter aerogenes 1 Proteus mirabilis 1

This table shows the bacterial strains isolated from the docu- mented infected ventricular catheters (n = 9). Coagulase ( - ) Staphy- lococcus constituted half of the isolates, and was found in con- junction with Staph. aureus in one case. The remainder of the in- fections were gram ( - ) rods. In 8/9 cases CSF pleocytosis (> 11 wbc/mm 3) was present at the time the culture for CSF was obtained (and eventually became positive for organism growth).

60

50

40

30

20

10

0 1 2 3 4 5 6 >6

Catheters Removed or Infected by Day

[] Infected

�9 Removed

Fig. 1. The lifespan for ventricular catheter duration is summarized in this graph. Nine catheters became infected, and are designated by top, light gray hatches. Over half of the catheters were removed before day five, but 6/9 infections occurred on days five and six

infection hazard as a function of time, assuming either a linear or exponential model. The Kaplan-Meier analysis accounted for cen- sored observations, since the patients exhibited a variable number of days of observation and most did not reach an infection endpoint.

Results

Infectious Complications

The admission diagnosis for each patient receiving a ventricular catheter is shown in Table 1, with the primary reasons for the catheter placement either pres- sure control or CSF drainage. The duration of catheter placement and number of infections is shown in Fig. 1. The majority of the catheters were censored (removed for non-infectious reasons) prior to day 5. A total of 9 catheters became infected, for an overall infection rate of 4.1%, predominantly on days 5 and 6. The pathogens isolated from CSF cultures for these cath-

C. G. Paramore and D. A. Turner: Relative Ventriculostomy Risks 8 l

eters are listed in Table 2; coagulase-negative Staphy- lococcus constituted half of the isolates. Gram ( - ) rods were grown from four cultures, while only one isolate contained Staph. aureus. In the majority of cases there was a CSF pleocytosis (> 11 wbc/mm 3) noted at the time of the first positive culture. Neither the patient's presenting diagnosis nor the location in which the ven- triculostomy was placed (i.e., intensive care unit) was significantly associated with the presence of CSF in- fection (data not shown).

Figure 2 demonstrates the proportion of uninfected catheters versus duration of catheter use. A steep de- crease in catheters remaining uninfected occurred at day 5, and another drop to 82.4% occurred the fol- lowing day. The daily hazard or risk of infection (as well as the simpler daily infection rate, as defined by Winfeld et aL 2~ is listed in Table 3 and plotted in Fig. 3. A steep increase in the hazard to 6.8% is noted between days 4 and 5, and this rises to 10.3% by day 6. Regres- sion analysis of the hazard indicated that these data were best fit by an exponential rather than linear func- tion (plotted as a semilogarithmic plot in the lower part of Fig. 3). Figure 3 (lower linear plot and upper semi- logarithmic plot) show the exponential regression line approximation to the infection risk, graphically de- picting the increasing daily infection risk with time.

Table 3. Daily Injection Hazard

Day Daily infection rate (%) Hazard (%)

1 0 0

2 0.53 0.59 3 0.67 0.74 4 0.83 0.96 5 4.5 6.78 6 6.9 10.26

The daily infection rate, defined as the ratio of the number of infections to the number of monitors in place on that day, is a relatively simple index of the incidence of infection which has been used by other authors 2~ The daily hazard function, or instantaneous rate of infection per day, is derived by the Kaplan-Meier method. It is a modification of the daily infection rate which compensates for the time interval being studied and the number of catheters removed for both infectious and non-infectious reasons. This com- parison demonstrates that the hazard gives more liberal estimates about the risk of infection, which may be important in drawing conclusions about the effect of time on infection rate. As no infections occurred in the ten catheters which remained beyond 6 days, no valid statistical conclusions about later time intervals may be made using this method. For interpretation see Fig. 3.

2

I �9

Z

100

90

00

i I i i i i i

1 2 3 4 5 6 7

C a t h e t e r D u r a t i o n (Days)

Fig. 2. A survival analysis showing cumulative, uninfected catheters remaining. A large number of catheters were treated as censored or incomplete data due to their removal at early time points. The cu- mulative infection-free group decreased sharply between days four and five, and then again between days 5 and 6, down to 82%. As no infections occurred after day six in this series, no conclusions can be inferred about the shape of the curve at later time points

1 /

0 ~ - . I

2 N cO

0

- 1

I I i /

I

I I I / #

/ f /,, / j . /

/1 / I z /

// 1 /

~ / - //

i . f J i i I I

1 2 S 4 5 6 7 C a t h e t e r D u r a t i o n (Days)

Fig. 3. The daily infection hazard (generated using the Kaplan-Mcier method) from Table 3 is shown plotted on both linear (upper) and logarithmic (lower) vertical axes. The daily hazard is the instanta- neous rate of infection for remaining catheters per day (taking into account those patients who are still at risk). By day five this rate was 5.6%, and climbed to 10.8% by day six. The increase in infectious hazard with time was shown to be statistically significant (regression slope test, p = 0.0163, with null hypothesis that the slope was flat or not different from zero). The dotted line ( - - ) represents an exponential function which has been fitted to the data to demonstrate the steep rise of the hazard after day four, which is a line function on the lower graph. Also shown in the lower graph ( - - ) are 95% confidence limits of the linear regression

82 C.G. Paramore and D. A. Turner: Relative Ventriculostomy Risks

Non-Infec t ious Complicat ions

Fifteen of the 253 catheter placements (5.6%) were noted to have non-infectious complications. Hemor- rhage occurred in 4 patients: two were asymptomatic, one patient suffered an acute subdural hematoma sec- ondary to a cortical vein laceration which required evacuation, and one patient had an ultimately fatal intraventricular hemorrhage. Thus, the rate of cata- strophic complication was 2/253, or roughly 1%. In 11 instances the catheter tip was noted to be outside of the lateral ventricle on a follow-up CT scan when the catheter demonstrated marginal CSF drainage. Sites of misplacement ranged from the internal capsule to the posterior clinoid (Table 4), but in no case was the patient noted to be symptomatic from the misplace- ment. All of these misplaced ventriculostomies were replaced with more appropriately targeted catheters.

The point at which the daily infection hazard reached the insertion risk was approximately day 5, suggesting that routine replacement at this point would balance these two risks. However, if two infections at either days 5 or 6 (out of the 10 total) were removed, the infection risk would have been consistent with a flat rate. Thus, determination of the infection risk is critically dependent on both the definition of infection (symptomatic with clinical ventriculitis versus positive CSF cultures) and the ascertainment as to when infec- tion occurs. In this study, all catheters were routinely cultured on removal and only occasionally were CSF cultures performed prior to withdrawal for clinical symptomatology.

Table 4. Non-Infectious Complications of Ventricular Catheter Placement

Complication Number

Subdural hematoma 1 Intracerebral hematoma 2 Intraventricular hemorrhage 1 Improper placement 11

The non-infectious complications of ventricular catheter place- ment are shown. Hemorrhage occurred in four patients: a subdural hematoma required surgical evacuation and an intraventricular hem- orrhage contributed to the demise of one patient. Catheter mis- placements varied from the deep gray near the lateral ventricle to the interhemispheric fissure and suprasellar cistern. There were no clinically evident signs related to these misplacements but all required a new catheter to be placed due to lack of CSF drainage.

Discussion

The overall infection rate for ventricular catheter placement (4.1%) lies in the range of published series, which include infection rates of up to 27% 1, 20, though a more current figure may be a cumulative infection rate of less than 50/05, 16, 19. The definition of CSF infection in this study was intentionally kept broad (positive CSF cultures rather than necessarily the pres- ence of symptomatic ventriculitis), so that any error would be on the side of overestimation of the true infection rate. Several authors, notably Mayhall et al. 1~

and Ohrstrom et al. 13 have also relied on positive CSF cultures for the diagnosis of ventriculitis. However Sundbarg e t a l ] 9 have argued that clinical symptoms and CSF pleocytosis are critical to establish the di- agnosis ofventriculitis. Sundbarg 19 reported 24 patients with positive CSF cultures but without clinical symp- toms or CSF pleocytosis that were given the diagnosis of"contamination"; 21 of these patients did not receive antibiotics. These patients apparently did not develop symptomatic ventriculitis or CSF infections, suggesting that there may have been contamination of CSF sam- ples 3. However, the incidence of postive CSF cultures in the absence of other clinical indicators of ventriculitis was sufficiently scarce in this series that all such patients were considered infected and treated aggressively. May- hall etal . 1~ reported that CSF pleocytosis (> 11 wbc/ mm 3) was significantly associated with positive CSF cultures, consistent with our current data.

Despite the low overall infection rate, our data dem- onstrate an exponentially increasing infection hazard with time. These results corroborate the prospective findings of Mayhall et al. 1~ and support the suspicion of several other authors 7' la, 15, 21. A possible weakness in the data is that CSF samples were not routinely withdrawn on a day to day basis, but only when a catheter was placed, withdrawn or when infection was clinically suspected. However, catheter withdrawal cul- tures were routine. As the majority of the catheters were replaced by day 5, most of the CSF sampling was also done by this time. It is unknown whether some of the infections could have been discovered earlier if CSF had been routinely sent each day. The infection rate prior to day 5 was exceedingly low; this fact somewhat mitigates the role of ascertainment bias in our results but could reflect the minimum time necessary for an insertion-related contamination to become clinically apparent. Interestingly, if as few as two of our infec- tions on days 5 and 6 were either not present or were actually infected earlier then our results would also be

c. G. Paramore and D. A. Turner: Relative Ventriculostomy Risks 83

consistent with a time-independent infection risk. Al-

though no infections occurred in the ten catheters which were left in for greater than 6 days (range 7-9 days), the Kaplan-Meier method does not allow interpreta- tion of data which extended beyond the last positive (i.e., infected) data point. Therefore we cannot draw

conclusions about the risk of infection at longer inter- vals.

Ohrstrom et al.~3 and more recently Winfield et al. a~

reported that the ventriculostomy infections in their series were evenly distributed over the entire interval the catheters were in place, suggesting a flat rather than an exponentially-increasing risk of infection as we have observed. In Winfield's series a~ the duration of catheter placement was more extended than in our series, but the majority of catheters had been removed by one week. However, the number of catheters infected re- mained relatively constant over a three week interval. Since the absolute number of infections per interval remained constant while the total number of catheters in place decreased rapidly, their conclusion that there was no increasing risk of infection with time must be viewed cautiously. Most studies on this subject, in- cluding our own, have too few long-term ventriculos- tomies to draw statistically valid conclusions about infection rates in catheters in place for more than one

week. This may be partially counteracted by using the actuarial hazard, as we have done, rather than a daily infection rate as performed by Winfield et al. 2~ since the hazard accounts explicitly for a decreasing number of catheters in the subsequent interval.

Non-infectious complications occurred in 5.6% of patients, and two-thirds of these were related to in- correct positioning of the catheter. Surprisingly, none of these patients suffered clinically detectible compli- cations related to the placement errors, despite two instances of placement into the internal capsule and one case in which the catheter was noted to be abutting the posterior clinoid. Only one patient in this series required an operation for subdural hematoma, while one suffered a catastrophic intraventricular hemor- rhage that was not amenable to surgical treatment,

leading to an overall hemorrhagic complication rate of 1.5%, similar to other series 12' 19

What conclusions can be drawn by comparing the relative risks of infectious and non-infectious (primarily placement) complications? By day 5, when the infec- tious hazard rate climbs to 6.8%, the risk of infection appears to be greater than the risk of replacing the catheter (5.6%), suggesting that ventricular catheters

should be replaced routinely at this time, in accordance with other reports 1~ 12. However, the infection rate for

ventriculostomy catheters remains substantial and pos- sibly may be reduced by other factors. The use of an- tibiotics for ventriculostomy catheters is not universally accepted13, 19 and no consensus exists on how to ad- minister such antibiotics 2' 21. In our study antibiotics were continued throughout the duration of ventricular cannulation, but antibiotics were rarely given prior to the procedure. However, if indeed a significant com- ponent of later catheter infection is initial contami- nation at the time of insertion then antibiotics should be administered prior to the placement of the catheter. Additionally, Mayhall et al. 1~ found a predictive value of only 0.538 for CSF pleocytosis in ventriculitis, but 8/9 of our infected population had a clear CSF pleo- cytosis by their criteria ( > 11 wbc/mm 3) at the time of diagnosis. A prospective study in which CSF was with- drawn daily for white blood count (and other) studies might reveal some factor or factors which would help the clinician to detect patients with potential ventri- culitis prior to the onset of clinical symptoms or positive cultures. Thus, further prospective data on antibiotic use either at the time of placement or continued for the duration of catheter presence and daily CSF sam- ples may delineate a more accurate infection hazard and may also prevent possible superinfection in these critically ill patients.

On the basis of this retrospective study we suggest that the risk of ventriculitis does increase with the length of catheter duration. A study of non-infectious complications has revealed that placement errors are relatively common, but in this series were not often associated with clinically detectible symptoms. Cath- eter replacement should be governed by careful analysis of both infectious and non-infectious complications, and should be done only when the risk of infection exceeds the risk of catheter insertion. Our data support the changing of ventricular catheters after a five day interval but further evaluation of prospective data is necessary for the validation of this practice as well as accompanying antibiotic administration.

Acknowledgements The authors were supported by grants from the B.S. Turner

Foundation, a Merit Review Research Award from the Department of Veterans Affairs (DAT) and NINDS Grant #RO1-NS29482 (OAT).

84 C.G. Paramore and D. A. Turner: Relative Ventriculostomy Risks

References

1. Aucoin PJ, Kotilainen HR, Gantz NM, Davidson R, Kellogg P, Stone B (1986) Intracranial pressure monitors: epidemiologic study of risk factors and infections. Am J Med 80:369-376

2. Blomstedt G (1985) Results of trimethoprim-sulfamethoxazole prophylaxis in ventriculostomy and shunting. J Neurosurg 62: 694-697

3. Buckwold FJ, Hand R, Hansebout RR (1977) Hospital-acquired bacterial meningitis in neurosurgical patients. J Neurosurg 46: 494-500

4. Kanter RK, Weiner LB (1984) Ventriculostomy-related infec- tions (letter). N Engl J Med 311:987

5. Friedman W, Vries J (1980) Percutaneous tunnel ventriculos- tomy - summary of 100 procedures. J Neurosurg 53:662 665

6. Gerner-Smidt P, Stenager E, Kock-Jensen C (1988) Treatment of ventriculostomy-related infections. Acta Neurochir (Wien) 91:47-49

7. James HE, Langfitt TW, Kumar VS, Ghostine SY (1977) Treat- ment of intracranial hypertension. Acta Neurochir (Wien) 36: 189-200

8. Kusske JA, Turner PT, Ojemann GA, Harris AB (1973) Ven- triculostomy for the treatment of acute hydrocephalus following subarachnoid hemorrhage. J Neurosurg 38:591-595

9. Lundberg N (1960) Continuous recording and control of ven- tricular fluid pressure in neurosurgicai practice. Acta Psych Neuro Scand [Suppl] 149:45-63

10. Mayhall CG, Archer NH, Archer-Lamb V, Spadora AC, Baggett JW, Ward JD, Narayan RK (1984) Ventriculostomy-related infections. N Engl J Med 310:553-559

11. Mollman HD, Rockswold GL, Ford SE (1988) A clinical com- parison of subarachnoid catheters to ventriculostomy and su- barachoid bolts: a prospective study. J Neurosurg 68:737-741

12. Narayan RK, Kishore PRS, Becker DP, Ward JD, Enas GG, Greenberg RP, DaSilva AD, Lipper MH, Choi SC, Mayhall CG, Lutz HA III, Young HF (1982) Intraeranial pressure: to monitor or not to monitor? J Neurosurg 56:650459

13. Ohrstrom JK, Skou JK, Ejlertsen T, Kosteljanetz M (1989) Infected ventriculostomy: bacteriology and treatment. Acta Neurochir (Wien) 100:67-69

14. Powell MP, Crockard HA (1985) Behavior of an extradural pressure monitor in clinical use. J Neurosurg 63:745-749

15. Rosner MJ, Becker DP (1976) ICP monitoring: complications and associated factors. Clin Neurosurg 23:494-519

16. Smith RW, Alksne JF (1976) Infections complicating the use of external ventrieulostomy. J Neurosurg 44:567-570

17. Stenager E, Gerner-Smidt P, Kock-Jensen C (1986) Ventricu- lostomy-related infections-an epidemiological study. Acta Neurochir (Wien) 83:20-23

18. Sundbarg G, Kjallquist A, Lundberg N, Ponten U (1972) Com- plications due to prolonged ventricular fluid pressure recording in clinical practice. In: Brock M, Dietz H (eds) Intracranial pressure. Springer, Berlin Heidelberg New York, pp 348-352

19. Sundbarg G, Nordstrom C, Soderstrom S (1988) Complications due to prolonged ventricular fluid pressure recording. Br J Neu- rosurg 2:485-495

20. Winfield JA, Rosenthal P, Kanter RK, Casella G (1993) Du- ration of intracranial pressure monitoring does not predict daily risk of infectious complications. Neurosurgery 33:424-431

21. Wyler AR, Kelly WA (1972) Use of antibiotics with external ventriculostomies. J Neurosurg 37:185-187

Correspondence: Christopher Paramore, M.D., Neurosurgery, Duke University Medical Center, Box 3807, Durham, NC 27710, U.S.A.