Object: Shunt infections and their management remain a clinically important problem in patients with hydrocephalus. The authors evaluated, in comparison with traditional parameters, C-reactive protein (CRP) in blood/serum (S-CRP) and in cerebrospinal fluid (CSF; C-CRP) for its power to identify and treat patients with infected shunts.
Methods: On 84 different occasions, CSF and blood samples from 59 children suspected of having shunt/CSF infections were obtained and evaluated. An infection was proven by a CSF culture in 35 of 84 evaluations. Values for S-CRP in infected individuals were higher than in noninfected ones (91.8 +/- 70.2 mg/L compared with 16.1 +/- 28.3 mg/L, p < 0.0001). The sensitivity of S-CRP testing was 97.1%, the specificity 73.5%, the negative predictive value 97.3%, and the receiver operating characteristic area 91.6%. The probability of shunt/CSF infection--provided that the S-CRP level was greater than 7 mg/L--rose from 41.7% (prevalence) to a posttest level of 72.3%. Specificity and posttest probability were 87.8 and 87.2%, respectively, if cases with other concurrent infections were excluded. The probability of missing a CSF/shunt infection at an S-CRP lower than 7 mg/L was 2.7%. All other diagnostic parameters did not yield useful test results. The rate of reinfection was elevated in patients in whom S-CRP levels were greater than 7 mg/L at the time of shunt reimplantation.
Conclusions: Analysis of these data suggests that the S-CRP level should be included in the first-line workup of patients with suspected shunt/CSF infection. It seems justified to avoid performing a shunt tap if S-CRP levels are less than 7 mg/L. A larger multicenter trial is necessary to confirm these promising diagnostic results and to deliver hard data concerning whether or not a normalized S-CRP level is a reliable indicator of successful antibiotic therapy and whether a shunt can be safely reimplanted.