Blood hemoglobin: a possible predictor of central venous catheter-related thrombosis in parenteral nutrition.

The biochemical records of 43 patients given nutritional support via a central venous catheter (CVC) for between 7 and 111 days (mean 28 days) without anticoagulant therapy were reviewed. On completion of the parenteral nutrition, phlebography was performed via the CVC. Patients developing CVC-related thrombosis (22/43, 51%) were assigned to group I and those without this phlebographic finding (21/43, 49%) to group II. The blood (B-) hemoglobin concentration before catheterization was higher in group I than in group II, the values being 121 +/- 16 (mean +/- SD) and 106 +/- 13 g/liter, respectively. An initial B-hemoglobin level above 127 g/liter showed a high positive accuracy of prediction for CVC-related thrombosis and on this basis 89% (8/9) of the patients were correctly assigned to group I, whereas an initial B-hemoglobin level below 111 g/liter showed a lower accuracy of prediction for a normal phlebographic status and on this basis 71% (12/17) of the patients were correctly assigned to group II. The specificity of the test criteria--B-hemoglobin greater than 127 and less than 111 g/liter--was high, with values of 95% (20/21) and 77% (17/22), respectively, whereas the sensitivity was low; only 36% (8/22) of group I had an initial B-hemoglobin above 127 g/liter and in group II 57% (12/21) recorded a value below 111 g/liter. It is concluded that if a patient has an initial B-hemoglobin concentration of more than 127 g/liter there is a high risk that CVC-related thrombosis will develop during parenteral nutrition lasting 1 wk or more.