Objective: To define the significance of various symptoms and signs in the diagnosis of ventriculoperitoneal shunt failure and infection.
Methods: The observations that form the basis of this study were made in the course of 2 multicenter, prospective, randomized, controlled clinical trials of technical aspects of ventriculoperitoneal shunt surgery-the Shunt Design Trial and the Endoscopic Shunt Insertion Trial. At registration, basic demographic and baseline clinical data were recorded. At scheduled follow-up visits 3 months and 1, 2, and 3 years after surgery and at unscheduled visits, the presence or absence of various symptoms or signs was recorded. At each visit, the neurosurgeon-investigator made a determination about whether the shunt had reached an end point: mechanical obstruction, infection, overdrainage, or loculation of the ventricular system. Observations at the last follow-up visit for each patient constituted the data for the current study. Sensitivities, specificities, and likelihood ratios were calculated for each symptom and sign as tests for shunt failure from any cause and for failure by infection. Decision trees were constructed to analyze the relationships of various symptoms and signs in the diagnosis of shunt failure and infection.
Results: Observations were available for analysis from 647 patient visits. A total of 248 shunts were judged to have failed (38%), and 55 were judged specifically to have failed by infection (8.5%). Bulging fontanel, fluid collection along the shunt, depressed level of consciousness, irritability, abdominal pain, nausea and vomiting, abnormal shunt pump test, accelerated head growth, and headache were strongly associated with shunt failure. Fever was strongly associated with shunt infection. Gross signs of wound infection were associated with shunt infection but were observed infrequently. Decision tree analysis confirmed the salience of bulging fontanel as a predictor of shunt failure. Fever and time since initial surgery were powerful predictors of shunt infection. Irritability emerged as an important observation in the identification of both shunt failure and shunt infection. Among children who underwent initial shunt insertion after 2 months of age, the absence of irritability, nausea/vomiting, and headache were powerful and generalizable predictors of the absence of shunt failure or infection.
Conclusions: Analysis of symptoms and signs of ventriculoperitoneal shunt complications can inform clinical judgment in the assessment of children with hydrocephalus.