Preoperative

In a prospective study of SO consecutive patients undergoing operation for ruptured abdominal aortic aneurysm, a coagulation screen was performed on admission to hospital. Twenty patients with either a platelet count < 100 x 109/1 or a prothrombin time > 1.5 times the control value had a mortality rate of 6.5 per cent (9.5 per cent confidence interval 45-8.5per cent); a further 23patients with normal screen results had a mortality rate of 9 per cent (9.5 per cent conjdence interval 0-20 per cent) (P < 0.001). Seven patients, of whom three died, did not have an admission coagulation screen performed. Patient age in the study group did not have independent statistical predictive power. This study indicates that coagulopathy at the time of admission predicts poor outcome in patients with ruptured aortic aneurysm. Current management strategies are inadequate for the treatment of these patients, who can be rapidly identified on admission by means of platelet and prothrombin counts.


Preoperative coagulopathy in ruptured abdominal aortic aneurysm predicts poor outcome
In a prospective study of SO consecutive patients undergoing operation for ruptured abdominal aortic aneurysm, a coagulation screen was performed on admission to hospital. Twenty patients with either a platelet count < 100 x 109/1 or a prothrombin time > 1.5 times the control value had a mortality rate of 6.5 per cent (9.5 per cent confidence interval 45-8.5per cent); a further 23patients with normal screen results had a mortality rate of 9 per cent (9.5 per cent conjdence interval 0-20 per cent) (P < 0.001). Seven patients, of whom three died, did not have an admission coagulation screen performed. Patient age in the study group did not have independent statistical predictive power. This study indicates that coagulopathy at the time of admission predicts poor outcome in patients with ruptured aortic aneurysm. Current management strategies are inadequate for the treatment of these patients, who can be rapidly identified on admission by means of platelet and prothrombin counts.
Ruptured abdominal aortic aneurysm (AAA ) continues t o be associated with a high hospital mortality rate. The inpatient mortality rate"3 ranges from 21 t o 70 (mean 49) per cent, despite the existence of specialized units employing rapid transport and resuscitation, and early surgical intervention by experienced personnel'. Current management of patients admitted t o hospital with a diagnosis of ruptured AAA consists of rapid resuscitation t o maintain systolic blood pressure above 80 mmHg, the use of pneumatic antishock garments and rapid transfer t o the operating theatre. Although several previous studies have attempted t o identify patients a t high risk of poor outcome, n o preoperative risk score has been shown t o be generally effective or widely a c~e p t e d~, ' .~ I' . Fa ctors associated with increased mortality rate include advanced age3.'ox'3, prolonged and profound h y p~t e n s i o n~. ' . '~. '~, preoperative renal impairment3.I4, cardiac failure".' 1.14,1' , chronic obstructive airway disease' and reduced level of consciousness".
Some studies have suggested that an association might exist between risk of death and pre-existing coagulopathy. Patients with evidence of disseminated intravascular coagulation (DIC ) a t presentation had mortality of 11 of 15, eight of nine, and four of four. The present study was undertaken t o find out whether a simple coagulation screen performed a t admission could discriminate patients with a high risk of perioperative death from those with a low risk, in such a way as t o provide a practical means for developing different management strategies for high-risk patients.

Patients and methods
A prospective study was carried out on 50 consecutive patients presenting to the accident and emergency department of a single hospital in whom a diagnosis of ruptured AAA had been made and who underwent operation. As soon as possible after admission, and before any blood or blood products had been transfused, blood samples were taken for a coagulation screen. This screen consisted of a platelet count and determination of the prothrombin time, activated partial thromboplastin time ( APTT), thrombin time and plasma fibrinogen concentration.
The platelet count was measured by an automated analyser and checked manually to exclude platelet clumping. The prothrombin time was measured by standard methodology and expressed as the international normalized ratio (INR). A platelet factor 111 reagent (partial thromboplastin 1 plus particulate activator was used to determine the APTT (Automated APTT; Organon Teknika, Cambridge, UK)". The thrombin time was measured using standard methodology that employs 3 units/ml calcium thrombin, adjusted to giveclotting times of9-11 s with normal plasma. The plasma fibrinogen concentration was measured using a modification of the Clauss technique".
Routine preoperative management consisted of establishing good venous access and, if the patient was hypotensive, infusion of crystalloids and colloids, and the application of a pneumatic antishock garment. When appropriate, patients were transfused with uncrossmatched 0-negative blood or cross-matched blood. Patients were transferred rapidly to the operating theatre. Transfusions of platelet concentrates and fresh frozen plasma were given during operation whenever a clinical diagnosis of established coagulopathy was made. All patients underwent emergency operation with aortic cross-clamping and insertion of a polytetrafluoroethylene vascular graft. Postoperative management was continued in the intensive care unit.
Patients were followed until death or discharge from hospital. A correlation was sought between the preoperative coagulation screen result and clinical outcome.
In the diagnosis of coagulation disorders, abnormal laboratory parameters can be demonstrated with simple tests such as platelet count, INR, APTT, thrombin time and fibrinogen levelz1. The main abnormalities are reflected by thrombocytopenia, prolonged prothrombin and thrombin times, and hypofibrinogenaemia2l. A fibrinogen level below 1.0 g/l (normal range 1.5-4 g/l) and a platelet count < 100 x 10y/l are regarded as virtually diagnostic, especially if associated with a generalized bleeding state". For the purposes of the present study an abnormal coagulation screen was defined as a platelet count < 100 x 109/1, INR, APTT and thrombin time > 1.5 times the control values, and a plasma fibrinogen level < 1.5 gjl.
Statistical analysis was performed using separate Wilcoxon rank sum tests for analysis of age, hospital stay and individual components of the coagulation screen, comparing patients who died with those who survived. Multiple logistic regression analysis was used to test the combined effect of the different factors on mortality. Comparison of mortality rate and length of hospital stay in the two groups (presence or absence ofcoagulopathy ) was by 1 ' analysis with Yates' correction.

Results
Fifty patients were entered into the study. As defined by the above criteria, 23 patients had normal and 20 abnormal admission coagulation screen results. Seven patients did not have a screen performed before blood transfusion was started or operation performed. It is felt that, in the clinical   (100) 15 of 43 (35) Values in parentheses are percentages. INR, international normalized ratio. *95 per cent confidence interval 0-20 per cent; t 9 5 per cent confidence interval 45-85 per cent circumstances surrounding admission of these critically ill patients, implementation of a research protocol in 86 per cent of cases is a reasonable achievement. Three of the seven patients who did not undergo screening died in the perioperative period, compared with 15 of 43 who did. There was no significant difference between these two groups, suggesting that patients in the unscreened group were not very different from those in the screened group. In addition, the unscreened group was further analysed by assuming a worst-case scenario in which all three patients who died would have had normal screen results had they been tested, and all four who survived would not (the probability of this scenario being true is about 0.007). In this case the mortality rate in the normal group would have been five of 26 (95 per cent confidence interval (c.i.) 7-39 per cent) compared with 13 of 24 in the abnormal group (95 per cent c.i. 33-74 per cent).
Overall, 18 patients died in the perioperative period; of these, 13 had an abnormal screen result, two a normal result and three no screen performed ( P < 0.001 for the difference in mortality rate between patients with normal and abnormal screen results; 95 per cent c.i. of the difference between rates 32-80 per cent). The cause and time of death are shown in Table I. As shown in Table 2, a low platelet count, raised INR and low plasma fibrinogen level were individually and collectively associated with a significantly increased risk of perioperative death. Thrombin time showed no significant difference between those who died and those who survived. A raised APTT lost its significant association with mortality rate when adjusted for platelet count or INR. Plasma fibrinogen level lost its significance when adjusted for platelet count. Patient age was not a significant independent variable.
There was a significant interaction ( P < 0.05) in the logistic regression between platelet count and INR, suggesting that survival was significantly worse when both were abnormal than would be expected from the effects of each separately. This supposition was confirmed clinically with a 100 per cent mortality rate in patients with both a platelet count < 100 x 109/1 and a prothrombin time > 1.5 times the control value ( Table 3). Death was less likely when only one of these parameters was abnormal. The length of hospital stay also showed a significant difference between the two groups. Patients with a normal preoperative coagulation screen result had a mean stay of 16 days, compared with 27 days for the seven with abnormal admission screen results who survived to leave hospital ( P < 0.001).

Discussion
Ruptured AAA remains a major and increasing cause of death23,24. The present study was performed in an attempt to find a rapid method for early identification of patients with ruptured aortic aneurysm who have a high risk of death and for whom current management strategies are not sufficient. Admission coagulation screen results were analysed on the basis of earlier observations that, among patients with a high mortality rate, DIC is a frequent In addition, coagulation screens can be performed with relative ease and speed, and could be a practical method for the rapid identification of high-risk patients.
The results indicate that the presence of coagulopathy on admission to the accident and emergency department is a strong predictor for subsequent failure of standard management regimens: only two of 23 patients with normal screen results died compared with 13 of 20 with abnormal results. Although the numbers in this study are too small to be certain, it is interesting to see that abnormality in platelet count plus INR is associated with a very high mortality rate. In patients with only one abnormality, the prognosis was not so poor.
Previous studies have demonstrated a n association between death from ruptured AAA and the presence of coagulopathy'6-18,25. The present study, however, shows that it is possible to identify patients at high risk of death on admission to the accident and emergency department by determining the INR and platelet count.
Patient age did not show independent significance for mortality. However, this observation may have been influenced by undeterminable preadmission biases; for example, very elderly patients may not have been referred by primary care physicians or other hospitals. None of the patients described had free intraperitoneal rupture of the AAA. Thus, the influence of established coagulopathy in this setting could not be evaluated. Prolonged preoperative hypotension may have contributed to the development of coagulopathy or to poor outcome. The time from rupture to arrival at the accident and emergency department could not be accurately determined. In all patients the time between arrival at hospital and entry to the operating theatre was less than 45 min, and for most it was less than 30 min.
The reason why coagulopathy predicts a poor outcome is not clear from the design or results of the present study. Although some of the deaths could clearly be linked to the presence of haemostatic failure, the link between the cerebrovascuiar accident 2 weeks after admission and coagulopathy on admission is obscure. The coagulopathy is possibly due to a consumptive disorder, reflecting the degree of tissue ischaemia before and during resuscitation, although this hypothesis will need to be proved in future studies. Some patients with asymptomatic AAA have low-grade D1C'6,26-28, and this may have contributed to the presence of coagulopathy. A simple dilutional coagulopathy arising during resuscitation before the blood sample was taken is improbable as no patient received more than 2 litres of colloid-crystalloid fluid intravenously before samples were taken. This volume is not associated with appreciable dilutional thrombocytopenia or hypofibrinogenaemia unless consumptive coagulopathy is also presentz9. None of the patients had a history of liver disease or was taking aspirin or anticoagulant medication before admission. Coagulopathy on admission may indicate a risk of excess bleeding at operation and can serve as a marker for ischaemic organ damage.
The present findings have important implications for the management of ruptured aortic aneurysm. Clearly, current management works well provided the patient is not in a high-risk group, but patients with pre-existing coagulopathy do badly, with an overall perioperative mortality rate of 65 per cent (95 per cent c.i. 45-85 per cent). This group can be identified rapidly on admission t o hospital: it should be possible to perform a platelet count and INR within 30 min of arrival. On the basis of these observations, there is a need for prospective studies assessing the ability of therapeutic intervention to improve outcome in high-risk patients with ruptured AAA. Possible therapeutic interventions could include aggressive haemodynamic support, with provision of plasma and platelet transfusions to correct coagulopathies, and the administration of protease inhibitors which have proven benefits in bleeding disorders as shown in cardiac" and liver3' transplant surgery. In addition, more extensive laboratory studies are required to analyse the mechanisms that contribute to the development and persistence of coagulopathy in these patients.