A new method for rationalising blood transfusion in paediatric surgery

Introduction  Two to three hundred transfusion complications are reported annually (1). Human variant CJD poses a relatively new and difficult concern with allogeneic blood transfusion. Cost (∼£100.00 per unit from donation to provision) is an additional consideration. With these and other considerations in mind the Association of Anaesthetists has published new guidelines for blood transfusion, which recommend a more cautious approach (2). A 5 year (July 1995–June 2000) retrospective review of transfusion requirements in children under 16 requiring debridement and split skin grafting was conducted in the North Trent regional burns unit, Northern General Hospital, Sheffield. In this 5 year period, 92 patients went to theatre a total of 107 times for debridement and grafting. 108 units of blood were cross-matched for 61 cases, with a total of 66 units transfused over 44 operations. Thirty-seven patients were over cross-matched with 6 under cross-matched. Analysis of the 5 year retrospective review reveals that with 10 g·dl−1 as an ideal target haemoglobin 8 patients were over transfused (to above normal range) and 19 were transfused more than necessary (to above 10 g·dl−1). Using 8 g·dl−1 as an ideal target haemoglobin the number of patients transfused more than necessary rises to 32. These results were analysed using a mathematical model based on 1st order kinetics with the equation dHb/Hb = dBL/EBV (where BL = blood loss and EBV = estimated blood volume) (3). This showed that without transfusion 17 of these patients would have had a Hb of above 8 g·dl−1 with 7 above 10 g·dl−1. Objectives  Using the above mathematical model to devise a computer aided means for 1 preoperative prediction of allowable blood loss before a threshold haemoglobin is reached 2 peroperative realtime prediction of haemoglobin concentration 3 peroperative realtime prediction of transfusion requirements 4 testing the model in a clinical setting Materials and Methods  Formulae developed were incorporated into a programme built using Microsoft Access 97. The programme facilitates the first 3 objectives outlined above. Ethical approval was obtained and a trial commenced in both a burns and spinal surgery setting to assess the how close computer predicted Hb results were to laboratory results. Key results  Thus far, 16 measurements taken during 2 burns and 5 spinal fusion operations show close correlation between predicted and tested Hb results. This was validated using the Bland Altman statistical method (see Fig. 1). Figure Figure 1. Bland Altman plot comparing predicted and measured Hb results following blood loss ranging from 5% to 23% of estimated circulating volume in 7 patients (16 measurements). The dotted lines indicate the mean difference ±2 standard deviations. Download figure to PowerPoint Conclusion  Computer aided prediction of Hb concentration concurrent with blood lost peroperatively could provide a useful cost effective tool in rationalising the use of allogeneic blood transfusion in paediatric surgical patients.