Sh u n t infection is a typical complication of shunt

Transcription

1 J Neurosurg Pediatrics 4: , 4: , 2009 Efficacy of antibiotic-impregnated shunt catheters in reducing shunt infection: data from the United Kingdom Shunt Registry Clinical article Hu g h K. Ri c h a r d s, Ph.D., 1,2 He l e n M. Se e l e y, Ph.D., 2 a n d Jo h n D. Pi c k a r d, F.Me d.sc i. 2 1 United Kingdom Shunt Registry, 2 Academic Neurosurgery Unit, Addenbrooke s Hospital, Cambridge, United Kingdom Object. In recent years CSF shunt catheters impregnated with rifampicin and clindamycin have been introduced to the United Kingdom (UK) market. These catheters have been shown to be effective in vitro against cultures of Staphylococcus epidermidis. The authors used data collected by the UK Shunt Registry to assess the efficacy of antibiotic-impregnated catheters (AICs) against shunt infection by using a matched-pair study design. Methods. The UK Shunt Registry contains data on nearly 33,000 CSF shunt-related procedures. The authors identified 1139 procedures in which impregnated catheters had been used, and accurate information was known about diagnosis, number of revisions, sex, and age in these cases. The database was ordered chronologically and searched forward and backward for cases with these same characteristics but involving conventional catheters. Matches were found for 994 procedures. Results. Among the 994 procedures in which AICs had been used, 30 shunts were subsequently revised because of shunt infection. Among the 994 controls, 47 were subsequently revised for infection (p = 0.048, chi-square test). Conclusions. The UK Shunt Registry does not collect data on causative organisms, and the surgeon is relied on entirely for the diagnosis of infection. However, with the large number of matched pairs evaluated, the authors attempted to reduce bias to a minimum. Their data suggest that AICs have the potential to significantly reduce shunt infections. (DOI: / PEDS09210) Ke y Wo r d s hydrocephalus antibiotic-impregnated shunt catheter shunt infection registry Sh u n t infection is a typical complication of shunt surgery and, after underdrainage, is the second most common reason given for shunt revision. Reported rates vary between 2 and 12%, with the age of the patient the most important determining factor. Initially the infection can be treated conservatively 8 but often requires removal of the shunt and replacement with an extraventricular drain, treatment of the infection with antibiotics, 1,8,16 and subsequent insertion of a new shunt. A prolonged stay in the hospital, typically 1 3 weeks, is often involved. There are distinct financial implications to a shunt infection the hospital stay, surgery, and implants probably cost between 20,000 and 30,000 as well as physical, emotional, and social costs. Hence, new technology capable of reducing the risk of shunt infection will probably be cost-effective. Abbreviations used in this paper: AIC = antibiotic-impregnated catheter; UK = United Kingdom. Rigid surgical protocols have been used to reduce shunt infections with some local success, but it is often not clear which aspects of the protocol are more effective in decreasing these infections. 9,10,21 Similarly, the efficacy of prophylactic antibiotics is unclear, with a variety of agents and routes being used with some reported success. 13,22,23 However, repeat infections are common, 16,17 and in the absence of a large clinical trial, the use of prophylactic antibiotics remains controversial 6 and may be effective only when the infection risk is high. 14 The leading cause of shunt infection is accepted as Staphylococcus epidermidis, with S. aureus and other bacteria and fungi (including yeasts) implicated as well. Shunt colonization with staphylococcus is difficult to eradicate once established because of the production of a biofilm. 5 In recent years CSF shunt catheters impregnated with rifampicin and clindamycin have been introduced to the UK market (Bactiseal, Codman and Shurtleff, Inc.). 389

2 H. K. Richards, H. M. Seeley, and J. D. Pickard These catheters have been shown to be effective against cultures of S. epidermidis and have a duration of action > 40 days in vitro. 3,4 The most compelling evidence for the efficacy or inefficacy of AICs would come from a randomized, doubleblind clinical trial. However, such a trial would be a major undertaking mainly because of the required sample size. In the absence of a large-scale clinical trial of AICs, several other experiences have been reported. Sciubba et al. 26 have revealed that changing from conventional (208 patients) to impregnated catheters (145 patients) significantly reduced the shunt infection rate in children from 12 to 1.4% at their hospital. A similar experience has been described by Pattavilakom et al. 20 in adult and pediatric patients the infection rate decreased from 6.5% in 551 procedures with standard catheters to 1.2% in 243 procedures with impregnated catheters. This difference was statistically significant. In a small study of 110 patients (two-thirds of whom were pediatric) in which randomization was applied, Govender et al. 12 found a significant (p = 0.04) reduction in the shunt infection rate with AICs, from 16.7 to 6%. A recent single-center study by Eymann et al. 11 compared 171 adults and 26 children who had received AICs between January 2002 and December 2006 with 98 adults and 22 children whose standard catheters had been inserted between January 1998 and December The overall infection rate decreased from 5.8 to 1% and was statistically significant (p = ). Interestingly, these authors included a cost-benefit analysis and concluded that if AICs can reduce infections by 50% when the initial infection is higher than 4%, then their use is cost-effective. In contrast, Ritz et al. 24 have conducted a study in adult and pediatric cohorts in whom the shunt infection rate (in cohorts of patients containing both adults and children) was 7.1% in those with conventional catheters (172 patients) and 6.9% in those with impregnated catheters (86 patients). Note, however, that this study did not include randomization. Catheters were chosen by the surgeon, which led to the use of AICs in patients who were significantly younger. Kan and Kestle 15 have described a study in which changing from conventional (80 patients) to impregnated catheters (80 patients) reduced the shunt infection rate in children from 8.8 to 5.0%. Similarly, Aryan et al. 2 have found a reduction from 15.2 to 3.1% in the infection rate in children on changing to impregnated catheters in a small study (46 standard and 32 impregnated catheters). The reduced infection rate in neither of these studies was statistically significant. These variations in results highlight the importance of a sufficient sample size. The overall shunt infection rate in the UK is estimated to be 5.2%. A trial intended to detect a reduction by one-half (at 80% power) would require the recruitment of ~ patients. In more vulnerable younger patients with an infection rate of 10%, a similar trial would require 870 patients. Amassing such numbers would necessitate cooperation between several centers, the establishment of a standard protocol, and considerable funding. No studies with patient numbers approaching these levels have been reported in the literature. The UK Shunt Registry is a resource that has collected data on large numbers of shunt procedures since May 1995 and is capable of generating the substantial data sets required; therefore, we have used data collected by the registry to assess the effectiveness of AICs against shunt infection in a matched-pair study design. Methods The UK Shunt Registry collects data on shunt and shunt-related procedures in the British Isles. All the major neurosurgery centers contribute. Data are gathered on a simple form, pads of which are available in the operating theaters. These forms are completed by theater staff and then returned to the registry. Every form is checked for data quality, and the information is entered into a database constructed with specialized registry software (Dendrite UK Ltd). The data are also validated by an auditor who periodically visits a contributing hospital and compares operating theater log books with data held on the UK Shunt Registry. We estimate that our overall national reporting rate from neurosurgery centers to the registry is 85%. We ask for the catalog and serial numbers of all catheters and valves used and, if the procedure is a shunt revision, the reason for revision. The reason for revision is purely intention-to-treat. In such a large database, it is impractical for the registry to confirm shunt infections or request data on causative organisms. The UK Shunt Registry contains data on > 33,000 CSF shunt-related procedures in > 21,700 patients. A preliminary examination of our data using logistic regression suggests that primary factors involved in shunt infection are patient age, sex, underlying pathology (that is, the original reason for shunting), and the number of revisions a patient has undergone. The timing of surgery and grade of the surgeon were found not to significantly contribute to the shunt infection risk. We performed an analysis by downloading data onto a spreadsheet (Microsoft Excel) and sorting by the patient identifier as the primary key and the date and time of the procedure as the secondary key. Custom software (written in Microsoft Visual Basic 6 and Microsoft Visual Basic for Applications) was used to identify procedures revised because of infection, and then, importantly, the previous procedure performed in that particular patient. This methodology allows identification of the catheter type that was subsequently revised for infection. To compare impregnated catheters with conventional catheters, a matched-pair study design was used. Proximal and distal catheters were identified by their catalog numbers, and we selected procedures in which both catheters were impregnated. Any procedure in which only 1 catheter was impregnated was excluded from the study. Procedures were included only if the follow-up period was at least 9 months. Data on the UK Shunt Registry indicate that 88% of shunt revisions for infection occur within 9 months of implantation. We identified 1139 procedures in which impregnated catheters had been used and there was accurate information on the underlying pathology, number of revisions (patients who had 5 revisions were pooled), patient sex, 390

3 Efficacy of antibiotic-impregnated shunt catheters and age. The database was ordered chronologically and searched forward and backward for the first occurrence of a patient matching all criteria but using conventional (unimpregnated) catheters. A leeway of 5% was allowed in matching the age of patients. If no match was found, a second search was made, allowing latitude in age of 10%. Any procedure not matched after a second pass was not included. Nine hundred ninety-four matched pairs were found. With 1988 patients, this data set exceeds the earlier discussed criteria for a trial. This design means that the 2 patient cohorts those with AICs and those with conventional catheters are a much closer match for known risk factors than can be achieved in a formal clinical trial. The demographic, revision, and diagnosis data shown in Table 1 are identical for both cohorts. The way that controls are chosen ensures that they are contemporary but still random, allowing a reduction in bias attributable to any unknown risk factors. The infection risks for AICs and standard catheters were calculated by identifying which catheters were subsequently revised due to infection, and the risks were compared using a chi-square test. Results The demographic composition of the matched-pair cohorts is shown in Table 1. Compared with the total of all shunt procedures in the UK in the equivalent period ( ), the matched-pair cohort is slightly biased toward pediatric neurosurgery, and the data also indicate that impregnated catheters are used relatively more often in primary shunt insertions. The median difference in age between the matched pairs was 0.6 years, and the median difference in the date of surgery was 0.3 years. Subsequent infections are shown in Table 2. In patients who received standard catheters, the infection risk was 4.7% (47 patients); in patients who received AICs, the infection risk was 3.0% (30 patients). This reduction is statistically significant (p = 0.048). The OR was 1.60 (95% CI ). Discussion It is generally accepted that the best way to evaluate a new procedure or device is by a formal prospective, randomized, double-blind clinical trial. In the case of impregnated catheters, however, blinding for the surgeon would be extremely difficult because of the catheters orange color. Such a trial would also prove a major undertaking because of the large number of patients required and the necessary collaboration between several neurosurgical centers. The advantages of a registry include the generation of a large data set and the belief held by each surgeon or center that patients are receiving the optimum treatment. The only ethical considerations concern the security of patient data. An analysis of registry data lacks the rigor of a randomized trial. Matched-pair studies are commonly used for large prospective population studies and have been used for an analysis of registry data. For example, data from the Eu- TABLE 1: Summary of demographic data in patients who underwent shunt procedures* Characteristic % sex F 49.6 M 50.4 age in yrs < > revision status primary insertion st revision 6.5 2nd revision 5.0 3rd revision 2.2 4th revision 7.8 primary diagnosis malformation 16.3 unspecified congenital 0.2 aqueductal stenosis 4.9 Dandy-Walker syndrome 1.1 Chiari malformation 3.3 spina bifida 6.8 other congenital 0.0 acquired hydrocephalus 73.2 cyst 3.0 colloid 1.2 arachnoid 1.6 unspecified 0.2 tumor 29.7 benign 15.9 malignant 13.2 unspecified 0.6 trauma 3.1 infection 7.5 meningitis 6.8 cerebral abscess 0.7 unspecified 0.0 posthemorrhage 29.9 perinatal IVH 14.7 AVM 1.5 SAH 13.4 unspecified 0.4 (continued) 391

4 H. K. Richards, H. M. Seeley, and J. D. Pickard TABLE 1: Summary of demographic data in patients who underwent shunt procedures* (continued) Characteristic % primary diagnosis idiopathic hydrocephalus 10.5 NPH 8.6 BIH 1.9 other 0.0 * AVM = arteriovenous malformation; BIH = benign intracranial hypertension (pseudotumor cerebri); IVH = intraventricular hemorrhage; NPH = normal-pressure hydrocephalus; SAH = subarachnoid hemorrhage. ropean Blood and Bone Marrow Transplant Registry have been used in a matched-pair evaluation of bone marrow and peripheral blood stem cell transplants. 18,27 The trauma registry of the German Trauma Society has been used in a matched-pair study of respiratory insufficiency with and without prehospital intubation, 25 and the International Pancreas Transplant Registry has performed a matched-pair comparison of transplant outcome using mycophenolate mofetil versus azathioprine. 13 In general, the pseudorandom selection of matched controls also adds to the robust nature of matched-pair studies. In the present study, the data were collected retrospectively, but the analysis was prospective. It is important to remove bias attributable to the use of AICs in patients with a perceived higher risk of subsequent shunt infection. This bias is a criticism of the study by Ritz et al. 24 described earlier. While the use of data matched for age, diagnosis, number of revisions, and sex will reduce this bias, it cannot be completely eliminated. However, one can argue that the inclusion of patients with a lower risk of shunt infection would reduce the likelihood of a significant result and might explain a lower reduction in infections than indicated by in vitro studies. The infection rate observed in controls (4.7%) is lower than expected if AICs are used in higher-risk patients. This finding may show that in our cohort AICs are used relatively more often in primary insertions. The registry does not collect microbiological data; the analysis is purely intention-to-treat, and it has been suggested that shunt blockage due to infection may be higher than recognized. 28 On the other hand, over-reporting of unconfirmed infection might have occurred, although this would apply equally to both arms of the study. Another source of bias may be the performance of neurosurgical centers or individual surgeons. For example, do more experienced surgeons use AICs? However, examination of the overall registry data suggests that there are no significant differences in shunt infection rates TABLE 2: Summary of data on shunt infections* Catheter Total No. Noninfected Shunts No. Requiring Subsequent Revision for Infection (%) AIC (3.0) standard (4.7) * p = 0.048; chi-square = either between neurosurgical centers or between different grades of surgeons. Even though there is a higher possibility of bias in the current study compared with a formal clinical trial, this bias is less than that in the studies involving a change to AICs and subsequent comparison with historical data. Such studies may lead to the introduction of confounding factors, often mistakenly called the Hawthorne effect (see discussions by Braunholtz et al. 7 and Parsons 19 ), whereby changes in protocol may confer an advantage independent of the catheters. Conclusions The cohort of patients (1988) analyzed in this study is larger than that in any previous investigation of antibioticimpregnated shunt catheters. We concluded that AICs can significantly reduce shunt infection from 4.7 to 3.0% in our cohort. While we accept that an analysis of retrospective data cannot be as robust as a prospective, randomized, double-blind clinical trial, the use of a matched-pair design is effective in reducing bias. Disclaimer Support for a research fellowship program has been established by industry, including Codman (the manufacturer of Bactiseal catheters), who receive data on the performance and usage of their own products. The UK Shunt Registry has complete independence over data collection, analysis, and publication. The registry has responsibilities to the Society of British Neurological Surgeons and the Medicines and Healthcare products Regulatory Agency. Acknowledgments The authors are grateful for the participation of centers (neurosurgical and pediatric) in the British Isles who perform shunt surgery. These centers not only supply our data, but also fund the registry. References 1. Arnell K, Enblad P, Wester T, Sjölin J: Treatment of cerebrospinal fluid shunt infections in children using systemic and intraventricular antibiotic therapy in combination with externalization of the ventricular catheter: efficacy in 34 consecutively treated infections. J Neurosurg 107: , Aryan HE, Meltzer HS, Park MS, Bennett RL, Jandial R, Levy ML: Initial experience with antibiotic-impregnated silicone catheters for shunting of cerebrospinal fluid in children. Childs Nerv Syst 21:56 61, Bayston R, Grove N, Siegel J, Lawellin D, Barsham S: Prevention of hydrocephalus shunt catheter colonisation in vitro by impregnation with anti microbials. J Neurol Neurosurg Psychiatry 52: , Bayston R, Lambert E: Duration of protective activity of cerebrospinal fluid shunt catheters impregnated with antimicrobial agents to prevent bacterial catheter-related infection. J Neurosurg 87: , Bayston R, Penny SR: Excessive production of mucoid substance in Staphylococcus SIIA: a possible factor in colonization of Holter shunts. Dev Med Child Neurol Suppl:25 28, Borgbjerg BM, Gjerris F, Albeck MJ, Børgesen SE: Risk of infection after cerebrospinal fluid shunt: an analysis of 884 firsttime shunts. Acta Neurochir (Wien) 136:1 7,

5 Efficacy of antibiotic-impregnated shunt catheters 7. Braunholtz DA, Edwards SJL, Lilford RJ: Are randomized clinical trials good for us (in the short term)? Evidence for a trial effect. J Clin Epidemiol 54: , Brown EM, Edwards RJ, Pople IK: Conservative management of patients with cerebrospinal fluid shunt infections. Neurosurgery 58: , Choksey MS, Malik IA: Zero tolerance to shunt infections: can it be achieved? J Neurol Neurosurg Psychiatry 75:87 91, Choux M, Genitori L, Lang D, Lena G: Shunt implantation: reducing the incidence of shunt infection. J Neurosurg 77: , Eymann R, Chehab S, Strowitzki M, Steudel WI, Keifer M: Clinical and economic consequences of antibiotic-impregnated cerebrospinal fluid shunt catheters. J Neurosurg Pediatr 1: , Govender ST, Nathoo N, van Dellen JR: Evaluation of an impregnated shunt system for the treatment of hydrocephalus. J Neurosurg 99: , Gruessner RWG, Sutherland DER, Drangstveit MB, Wrenshall L, Humar A, Gruessner AC: Mycophenolate mofetil in pancreas transplantation. Transplantation 66: , Haines SJ, Walters BC: Antibiotic prophylaxis cerebrospinal fluid shunts: a metanalyis. Neurosurgery 34:87 93, Kan P, Kestle J: Lack of efficacy of antibiotic-impregnated shunt systems in preventing shunt infections in children. Childs Nerv Syst 23: , Kestle JRW, Garton HJL, Whitehead WE, Drake JM, Kulkarni AV, Cochrane DD, et al: Management of shunt infections: a multicenter pilot study. J Neurosurg 105: , Kulkarni AV, Rabin D, Lamberti-Pasculli M, Drake JM: Repeat cerebrospinal fluid shunt infection in children. Pediatr Neurosurg 35:66 71, Majolino I, Pearce R, Taghipour G, Goldstone AHJ: Peripheral-blood stem-cell transplantation in Hodgkin s and non- Hodgkin s lymphomas: a new matched-pair analysis of the European Group for Blood and Marrow Transplantation Registry data. J Clin Oncol 15: , Parsons HM: What happened at Hawthorne? Science 183: , Pattavilakom A, Xenos C, Bradfield O, Danks RA: Reduction in shunt infection using antibiotic impregnated CSF shunt catheters: an Australian prospective study. J Clin Neurosci 14: , Pople IK, Bayston R, Hayward R: Infection of cerebrospinal fluid shunts in infants: a study of etiological factors. J Neurosurg 77:29 36, Ragel BT, Brown SR, Schmidt RH: Surgical shunt infection: significant reduction when using intraventricular and systemic antibiotic agents. J Neurosurg 105: , Ratilal B, Costa J, Sampaio C: Antibiotic prophylaxis for surgical introduction of intracranial ventricular shunts: a systematic review. J Neurosurg Pediatr 1:48 56, Ritz R, Roser F, Morgalla M, Dietz K, Tatagiba M, Will BE: Do antibiotic-impregnated shunts in hydrocephalus therapy reduce the risk of infection? An observational study in 258 patients. BMC Infect Dis 7:38, Ruchholtz S, Waydas C, Ose C, Lewan U, Nast-Kolb D: Prehospital intubation in severe thoracic trauma without respiratory insufficiency: a matched-pair analysis based on the Trauma Registry of the German Trauma Society. J Trauma 52: , Sciubba DM, Stuart RM, McGirt J, Woodworth GF, Samdani A, Carson B, et al: Effect of antibiotic-impregnated shunt catheters in decreasing the incidence of shunt infection in the treatment of hydrocephalus. J Neurosurg 103: , Sirohi B, Powles R, Kulkarni S, Rudin C, Frassoni F, Bacigalupo A, et al: Reassessing autotransplantation for acute myeloid leukaemia in first remission a matched pair analysis of autologous marrow vs peripheral blood stem cells. Bone Marrow Transplant 33: , Walters BC, Hoffman HJ, Hendrick EB, Humphreys RP: Cerebrospinal fluid shunt infection. Influences on initial management and subsequent outcome. J Neurosurg 60: , 1984 Manuscript submitted August 11, Accepted April 28, Address correspondence to: Hugh K. Richards, Ph.D., UK Shunt Registry, Academic Neurosurgery Unit, Box 167, Addenbrooke s Hospital, Cambridge CB2 0QQ, United Kingdom. hkr10@ medschl.cam.ac.uk. 393