Haemorrhage following pancreatoduodenectomy: the importance of surgery.

Koukoutsis et al. [4] subsequently analyzed risk factors in predicting post-operative haemorrhage, particularly secondary haemorrhage, yet they do not provide the raw data to allow an in-depth analysis of these factors, for example the number of patients who had a sentinel bleed but did not have a post-operative bleed, or the total number of patients in the series who had an anastomotic leak. This, combined with the question over the grouping of these patients and the lack of multivariate analysis (although the small numbers limit its use) questions how much can be extracted from these results. However, importantly, the sign of a herald bleed should prompt further investigation, preferably a contrast-enhanced CT scan with CT angiography [4, 6] . Any collections or uncontrolled anastomotic leaks should be promptly drained, while any pseudo-aneurysms should be embolized. Busch et al. [6] also recommend angiography in those with massive intra-abdominal haemorrhage, as operating on these patients can be a formidable undertaking with a high mortality [4, 7] ; yet data from their own unit indicated a bias toward surgical intervention [7] . Even if embolization is successful, surgery is still likely to be required to deal with the underlying cause (anastomotic leak), but it is suggested that this is more feasible in stable patients [6] . Koukoutsis et al. [4] employed a combination of techniques to deal with post-operative bleeding and although the numbers are too small to make definitive conclusions, it is worth noting that the success of emboWith the development of specialist centres, mortality following pancreatoduodenectomy has fallen to below 5%, yet morbidity remains between 30 and 50% [1–3] . Major haemorrhage following pancreatoduodenectomy makes a significant contribution to this ongoing morbidity and mortality [3] . Significant haemorrhage following pancreatoduodenectomy occurs in less than 10% [2, 3] of patients, thus identifying risk factors and determining the subsequent optimal management can be difficult unless large series are analyzed. In this issue of Digestive Surgery, two such series [4, 5] are presented. Koukoutsis et al. [4] report on a large series (n = 362) of consecutive pancreatoduodenectomies over a 4-year period. Haemorrhage occurred in 8.8% (n = 32) of patients, with a 47% (n = 15) associated mortality. The authors [4] divided post-operative haemorrhage into ‘primary haemorrhage’, defined as occurring within the first week and usually due to technical issues involving haemostasis, and ‘secondary haemorrhage’, defined as occurring later due to sepsis and ruptured pseudo-aneurysms. (More correctly perhaps the authors should have defined primary haemorrhage as that occurring within 24 h, and then early and late secondary – but we shall keep to their definition). Primary haemorrhage occurred in 18 of the 32 patients who suffered post-operative bleeding and, strikingly, 6 of these 18 patients died. Of the 18 patients in this group, 5 suffered a pancreatic anastomotic leak, 7 an intra-abdominal abscess and 6 had evidence of sepsis. Published online: July 26, 2005