measured to assess the acute-phase response. Clinical outcome was determined using mortality, morbidity and the incidence of the systemic in¯ammatory response syndrome SIRS). Results: There was no difference between the ANH and control group in serial measurements of median range) white cell count maximum at 2 days after operation: 11´9 7´7±21´4) versus 10´3 7´8±20´6) 3 10 9 l ±1 ; P = 0´25), serum C-reactive protein level maximum at 3 days: 150 1±274) versus 169 7± 238) mg ml ±1 ; P = 0´76), interleukin 6 level maximum at 6 h: 142 32±793) versus 105 29±509) pg ml ±1 ; P = 0´89), total antioxidant capacity lowest at 1 h: 0´83 0´67±1´22) versus 0´83 0´68±1´23) mmol l ±1 ; P = 0´45) or urinary albumin/crea- tinine ratio maximum at 30 min after clamp release: 41 2±923) versus 124 4±376) mg ml ±1 ; P = 0´10). SIRS was observed in ten of 16 patients having ANH and in 11 of 18 control patients P = 0´99). There was no signi®cant difference in mortality and morbidity between the groups. Similarly, there was no difference in median range) blood loss ANH 1800 400± 12 000) ml versus control 1600 500±7500) ml; P = 0´55), use of cell salvage 600 0±4740) versus 520 0±2420) ml; P = 0´60) or heterologous blood transfusion 2 0±32) versus 2 0±9) units; P = 0´68). Conclusion: In the setting of a randomized controlled trial ANH added no additional bene®t, when used in combination with cell salvage, in reducing the requirements for heterologous blood transfusion, and made no impact on systemic in¯ammatory response and clinical outcome after AAA repair. Aneurysm-related mortality during late follow-up after endovascular aneurysm repair of infrarenal aorta S.R.Vallabhaneni,P.L.Harris,G.L.Gilling-Smithand R.LaheijonbehalfoftheEUROSTARcollaborators Royal Liverpool University Hospital, Liverpool, UK Background: Aneurysm-related mortality ARM) accounts for around 1´5 per cent of all deaths following open aneurysm repair. The incidence of ARM following endovascular aneurysm repair EVAR) is unknown. The aim was to examine all causes of death, including ARM, during late follow-up after EVAR. ARM was de®ned as death resulting directly from rupture of the repaired aneurysm or another complication of the aneurysm, more than 30 days after repair, or death within 30 days of a secondary intervention undertaken solely to rectify a complication of repair. Methods: Preoperative and follow-up data on 2194 patients from 88 European centres were collected prospectively on to a database. Survival up to 48 months after EVAR was analysed by means of Kaplan±Meier survival analysis. The causes of death during this period were noted. Results: There were 161 deaths between 1 and 48 months after EVAR. The cumulative rate of secondary intervention for this cohort at 4 years was 33 per cent. The causes of death were: cardiac 28´6 per cent, malignancy 18´6 per cent, cerebrovas- cular 6´8 per cent, respiratory 3´1 per cent, renal 1´8 per cent, other 22´3 per cent and ARM 11´8 per cent. There were 19 deaths from aneurysm-related causes. Nine patients died following proven rupture of the aneurysm, three died from presumed rupture of the aneurysm and a further seven died following late conversion three patients), graft sepsis two) and secondary intervention two). Sudden death of uncertain cause occurred in ten patients in whom rupture of AAA was a possibility. Conclusion: Non-aneurysm-related causes of death were comparable to those in published reports of survival after open repair. However, the proportion of aneurysm-related deaths 11´8 per cent) was appreciably higher than that reported after open repair. These results may re¯ect the learning curve experience of the teams involved in the study, but continued caution is advisable regarding expectations of outcome following EVAR. Autologous or homologous transfusion in aortic surgery: randomized trial F.Torella,J.C.L.Wong,S.L.HaynesandC.N.McCollum South Manchester University Hospital, Manchester, UK Background: Aortic surgery often requires blood transfusion, which may cause complications and postoperative infection. Autologous transfusion was evaluated in a multicentre clinical trial. Methods: Some 145 patients undergoing elective aortic surgery in eight hospitals were randomized to either `homologous' or `autologous' transfusion, a combination of acute normovolae- mic haemodilution ANH) and intraoperative cell salvage. Homologous blood was administered when the haemoglobin concentration fell below 8 g dl ±1 . Results: Median interquartile range i.q.r.)) blood loss was 668 400±862) ml or 17 10±24) per cent of blood volume in aortobifemoral bypass, and 1120 765±1700) ml or 24 17±36) per cent in aneurysm repair P < 0´001). Autologous transfusion reduced homologous blood requirements from a median i.q.r.) of 2 0±4) units to 0 0±2) units P = 0´008). Independent predictors of blood transfusion were homologous transfusion strategy odds ratio OR) 2´3 95 per cent con®dence interval 1´1±5´0); P = 0´03), low preoperative haemoglobin concentra- tion OR 3´7 1´7±8´2); P < 0´001), prolonged surgery OR 2´1 1´0±4´8); P = 0´05) and blood loss OR 3´0 1´4±6´5); P = 0´007). Patients with a preoperative haemoglobin concentration greater than 13´5 g dl ±1 and who lost less than 20 per cent of their blood volume rarely required transfusion. There was no signi®cant difference between the groups in terms of morbid- ity, mortality and postoperative hospital stay. Conclusion: Autologous transfusion reduced the need for homologous blood in aortic surgery, but was useful only in patients with low haemoglobin levels or when blood loss exceeded 20 per cent of the blood volume. ANH alone is indicated for patients undergoing aortobifemoral bypass and in those with a higher haemoglobin level and blood volume. 604 VSSGBI abstracts British Journal of Surgery 2001, 88, 598±622 www.bjs.co.uk ã 2001 Blackwell Science Ltd