Diagnostic Accuracy of Clinical Signs and Biochemical Parameters for External Ventricular CSF Catheter-Associated Infection

Abstract

Background and objectives: Few prospective well-designed diagnostic accuracy studies have been performed to study the parameters of infection in patients suspected for external ventricular catheter-associated infection. Our objective was to analyze the diagnostic accuracy of clinical characteristics and biochemical and microbiological parameters in diagnosing external ventricular CSF catheter-associated infection.

Methods: From 2014 to 2017, we performed a single-center cohort study in consecutive patients at the intensive care unit who required an external ventricular CSF catheter in the Hague, the Netherlands. CSF was sampled and analyzed daily. Ventricular catheter-associated infection was defined according to the 2017 Infectious Diseases Society of America's Clinical Practice Guidelines. We compared clinical characteristics and biochemical parameters between patients with and without infection from 3 days before to 3 days after the day the CSF sample was collected that grew bacteria.

Results: A total of 103 patients were included of whom 15 developed a catheter-associated infection (15%). The median day cultures were positive was 3 days after CSF collection (interquartile range [IQR] +2 to +4). On day 0, none of the tests could differentiate between patients with and without infection. The CSF leukocyte count was increased in patients with ventricular catheter-associated infection as compared with patients without on days +2 and +3. The difference was most prominent on day +2 (1,703 × 10⁶/L [IQR 480-6,296] vs 80 × 10⁶/L [IQR 27-251]; p < 0.001; area under the curve [AUC] 0.87 [95% confidence interval (CI) 0.71-1.00]). Sensitivity for the CSF leukocyte count at a cutoff level >1,000 × 10⁶/L was 67% (95% CI 30-93), and specificity was 100% (95% CI 90-100); the positive predictive value was 100%, and the negative predictive value was 92% (95% CI 83-97). The percentage of polymorphonuclear cells (PMNs) was higher in patients with infection on days +1 and +2 (day +2 89% [IQR 78-94] vs 59% [IQR 39-75]; p < 0.01; AUC 0.91 [95% CI 0.81-1.0]).

Discussion: An elevated CSF leukocyte count and increased percentage of PMNs are the strongest indicators for external catheter-associated infections on the days before culture positivity. New CSF markers of drain-associated infection should be studied to enable earlier diagnosis and treatment in patients with an infection and reduce antibiotic treatment in those with no infection.

Classification of evidence: This study provides Class I evidence that in individuals requiring an external ventricular CSF catheter, an elevated CSF leukocyte count and an increased percentage of PMNs are the strongest indicators of catheter-associated infections in the days before CSF culture positivity.

Figure 1. Infection Occurrence Over Days

Figure 2. CSF Measures Over Time (Leukocyte Count, % of Polymorphonuclear Cells, Protein Concentration, and Cell Index)

Median results and IQR for leukocyte count in CSF (A), percentage of polymorphonuclear cells (B), total protein concentration in CSF (C), and cell index (D) on the 3 days before infection and the first 3 days of infection. Significant differences are indicated with *. IQR = interquartile range.

Figure 3. CSF Measures Over Time (Glucose Concentration and CSF/Blood Ratio, Lactate Concentration and Ratio)

Median results and IQR for glucose concentration in CSF (A), CSF-to-blood glucose ratio (B), lactate concentration in CSF (C), and CSF-to-blood lactate ratio in blood (D) on the 3 days before infection and the first 3 days of infection. Significant differences are indicated with *. IQR = interquartile range.