Object: The physiological reaction of CSF white blood cells (WBCs) over the course of treating a shunt infection is undefined. The authors speculated that the CSF WBC count varies with different infecting organisms in peak level and differential percentage of polymorphonuclear (PMN) leukocytes, lymphocytes, monocytes, and eosinophils. The authors hope to identify clinically useful trends in the progression of CSF WBCs by analyzing a large group of patients with successfully treated shunt infections.
Methods: The authors reviewed 105 successfully treated cases of shunt infections at Riley Hospital for Children. The study dates ranged from 2000 to 2004; this represented a period prior to the routine use of antibiotic-impregnated shunt catheters. They analyzed the following organisms: coagulase-negative staphylococci, Staphylococcus aureus, Propionibacterium acnes, Streptococcal species, and gram-negative organisms. The initial CSF sample at diagnosis was analyzed, as were levels over 14 days of treatment. Model fitting was performed to generate curves for the expected progression of the WBC counts and the differential PMN leukocytes, lymphocyte, monocyte, and eosinophil percentages.
Results: Gram-negative organisms resulted in a higher initial (p = 0.03) and peak WBC count with a greater differential of PMN leukocytes compared with other organisms. Propionibacterium acnes infections were associated with a significantly lower WBC count and PMN leukocytes percentage (p = 0.02) and higher eosinophil percentage (p = 0.002) than other organisms. The pattern progression of the CSF WBC count and differential percentages was consistent for all infections. There was an initial predominance of PMN leukocytes, followed by a delayed peak of lymphocytes, monocytes, and eosinophils over a 14-day course. All values trended toward zero over the treatment course.
Conclusions: The initial and peak levels of CSF WBCs vary with the infecting organisms. The CSF cell counts showed a predictable pattern during the treatment of shunt infection. These trends may be useful to the physician in clinical decision making, although there is a wide range of variability.