Variations of the blood cell count in infection.

Purpose. - Leukocytosis is considered as an argument for infection. We have compared leukocytosis with the other data available with the white blood cell count. Methods. - White blood cell counts obtained from 187 patients (age: 18 to 81 years), admitted to an emergency room for abdominal pain, were analysed using an automate (Bayer Technicon H2, Dublin, Ireland), and compared with matched healthy subjects. The patients were classified into two groups: diseases of the biliary or of the urinary tract, and subsequently in two subgroups: infectious diseases (angiocholitis, cholecystis, pyelonephritis) or non-infectious diseases (hepatic and nephritic colic). Results. - Leukocytes and neutrophils were significantly increased, and eosinophils significantly decreased in all subjects by comparison with controls. These abnormalities were more important in infected patients. Lymphocytes were significantly decreased in infectious disease. For predicting infection, sensitivity and specificity of leukocytosis (> 1,000/mm(3)) were respectively 66% and 56%, while that of eosinopenia (< 100/mm(3)) were respectively 91% and 38%, and that of lymphopenia (< 1,2001 mm(3)) respectively 58% and 73%. The probability of infection was less than 3% when neutrophils were less than 7, 000/mm(3), and eosinophils and lymphocytes respectively more than 100/mm(3) and 1,2001 mm(3). Conclusion. - This study shows that leukocytosis, eosinopenia or lymphopenia are poor indicators of infection, when considered alone or in combination. However, eosinopenia and lymphopenia appear as better criteria of infection than leukocytosis. A detailed analysis of the white blood cell count allows the exclusion of infection with an acceptable risk of error (C) 2002 Editions scientifiques et medicales Elsevier SAS.