the actions needed to prevent harm” (69). The nine unrelated action items involved devices such as knee prostheses, airways, and pacemakers that use some of the terms in the search but are not CSF shunt devices. The five action items involved risk of fracture of a right angle connector leading to a recall of unimplanted connectors (from a batch of 4,600 units), debris in metal connectors leading to a recall of 156 units, abnormally high operating pressure of valve leading to recall of all unimplanted units (number not specified), mislabeled closing pressure on valves leading to recall of unimplanted valves (from a total of 60,760 distributed), and the distribution of ventricular catheters without holes in them. The two most recent actions were in 1992 and 2004.

Data Mining Techniques

We applied disproportionality analysis techniques (see Appendix D) to the shunt subset of the MAUDE database. Shunt complications in the database were identified by searching the text of the narrative descriptions of adverse events in the database (the details are given in Appendix D of the main report). This analysis identified only two associations between specific manufacturers and adverse event reports. Table E.7 shows the results of the analysis, identifying an excess of catheter disconnections associated with products of Company A and an excess of catheter migration associated with products of Company B.

We could not identify any action such as a recall, safety alert, or publication possibly related to the migration events identified for Company B. In the case of company A, concern was expressed in the literature regarding fracture (included in our definition of disconnection) in papers published in 1992 and 1995 (60), (70), (71). These papers deal with predominantly pediatric populations. A change in the formulation of the catheter was made by the company in response to concerns about an excessive number of catheter fractures. As the public data in the MDR and MAUDE databases does not include the age of the patient, we could not determine if children were overrepresented in the adverse event reports. No author has

TABLE E.7 Disproportionality Analysis of CSF Shunt Events in MAUDE Database



















NOTE: N = number of events reported, E = number of events expected in the absence of association, RR = raw relative risk estimate (N/E), EBGM (Empirical Bayes Geometric Mean) is an adjusted relative risk estimate. For details see Appendix C.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement