European Journal of Vascular & Endovascular Surgery
Volume 36, Issue 6 , Pages 637-645, December 2008

Predicting Risk in Elective Abdominal Aortic Aneurysm Repair: A Systematic Review of Current Evidence

St George's Vascular Institute, 4th Floor, St James Wing, St George's Hospital, London SW17 0QT, UK

Received 11 July 2008; accepted 27 August 2008. published online 15 October 2008.

Article Outline

Abstract 

Objective

To examine and compare existing pre-operative risk prediction methods for elective abdominal aortic aneurysm (AAA) repair.

Design

Systematic review.

Methods

Medline, EMBASE and the Cochrane library were searched for articles that related to risk prediction models used for elective AAA repair.

Results

680 abstracts were reviewed and after exclusions 28 articles encompassing 10 risk models were identified. The most frequently studied of these were the Glasgow Aneurysm Score (GAS), the Physiological and Operative Severity Score for enUmeration of Mortality (POSSUM) predictor equation and the Vascular Biochemistry and Haematology Outcome Model (VBHOM). All models had strengths and weaknesses and some had unique features which were identified and discussed.

Conclusion

The GAS appeared to be the most useful and consistently validated score at present for open repair. Other systems were either not validated fully or were not consistently accurate. Some had significant drawbacks which appeared to severely limit their clinical application. Recent work has shown that no scores consistently predicted the risk associated with endovascular aneurysm repair (EVAR). Pre-operative risk stratification is a vital component of modern surgical practice, and we propose the need for a comprehensive new risk scoring method for AAA repair incorporating anatomical and physiological data.

Keywords: Aortic aneurysm, Abdominal, Mortality, Risk adjustment, Endovascular aneurysm repair

 

Back to Article Outline

Introduction 

The rupture of previously undetected abdominal aortic aneurysms (AAAs) is responsible for a significant number of deaths in the UK, with an overall mortality of over 80%. This has not improved in the last fifty years despite a reduction in the risk associated with elective repair in this period.1, 2, 3 Older patients with greater co-morbidity are increasingly undergoing elective repair to prevent rupture. The risk and potential benefit of such procedures must be considered to determine which should be approached with greater caution.

Across many surgical specialties mathematical models have been used to aid surgeons to plan operations. Vascular-specific scoring systems have been derived with AAA repair forming an area of particular interest. At present there is no consensus as to which scoring system, if any, should be used routinely. Therefore, the current evidence for the use of risk prediction methods for both open and endovascular techniques was reviewed.

Back to Article Outline

Methods 

Medline, EMBASE and the Cochrane Collection were interrogated. The keywords used were “abdominal aortic aneurysm”, and “risk” and/or “score”. Abstracts of all articles that proposed new scoring systems or externally validated existing systems were reviewed. Further potentially useful articles were identified through scrutiny of references. Articles were excluded if they examined risk associated with individual factors in isolation and did not produce a predictive model. Those primarily concerned with thoraco-abdominal aneurysms or aortic dissection were also not considered. This review was concerned with risk prediction of elective AAA repairs, so articles exclusively examining ruptured AAA repair were excluded. Scores requiring intra-operative or intensive care data were not considered unless a modified “physiology-only” version existed, as they could not be used to predict an individual's risk pre-operatively. Articles that described systems that were not used subsequently or were derived from small populations were excluded as they could not be compared meaningfully to other established scores.

The main comparator was the area under the curve (AUC) of a plotted receiver–operator characteristic (ROC) curve, which indicated discriminative power. Values from 0.70 to 0.50 denoted progressively poorer predictive power than chance alone, with values over 0.80 indicating reliable accuracy. The concordance statistic or c-statistic was equivalent to the AUC of an ROC. Another aspect of usefulness was the calibration of a model to the population it was being validated against. This was measured using the Hosmer–Lemeshow test for goodness-of-fit expressed as statistical value, degrees of freedom (df) and p-value (p=x).

Of 680 potential articles identified in the searches, 602 were excluded according to the stated criteria as not being relevant from a review of the abstract. The remaining 78 articles underwent detailed review. At this stage 46 articles were excluded as the models described required peri-operative details, described scores for ruptured AAA repair, thoracic aneurysms or aortic dissection. Of the 32 remaining articles, four more were excluded because they described models that had very small sample size, were not validated in any meaningful way or had never been studied (Table 1). After exclusions 28 articles remained encompassing 10 risk prediction models (Table 2). These were reviewed in depth.

Table 1. QUORUM flow-chart for the search process
Table 2. Summary of items used in each score
AgeSexASAIHDMICCFECG abn.HTCVDRFRespDMBld loss/HbU+EWBCOther
GAS• • ••
POSSUM• • ••• •• •••GCS
VBHOM• • •••White cell count
E-PASS• •• • Performance index
Leiden/modified Leiden•• ••• •• Centre-specific weighting
Eagle• • • • Myocardial perfusion
Vanzetto• •• •• • Myocardial perfusion
Modified Lee/combined prognostic index • • ••• Protective effect of statins and beta-blockers
Australian National Audit Model••• • Anatomical risk factors
SVS co-morbidity score• • • •

(•)=Items common to each score-but note that the table is not comprehensive. ASA – American Society of Anaesthetists score; IHD – myocardial disease or angina; MI – history of myocardial infarction; CCF – congestive cardiac failure; HT – hypertension; CVD – cerebrovascular disease; RF – renal failure; Resp – respiratory disease; DM – diabetes mellitus; U+E – urea and electrolytes; WBC – white cell count.

Back to Article Outline

Results 

Models for predicting outcome following elective open aortic aneurysm repair 

Glasgow Aneurysm Score 

The GAS was constructed using logistic regression in 1994 using a population of 500 patients undergoing AAA repair between 1980 and 1990. It identified pre-operative shock, myocardial dysfunction, renal impairment and cerebrovascular disease as significant factors in determining post-operative outcome.4 It was validated successfully by the same group shortly afterwards, confirming the accuracy of this method of predicting the immediate outcome of elective and emergency open AAA repair.5 Further work in 2003 validated the GAS using data from 403 patients operated on in a single hospital, and then again on a separate occasion using the Finnvasc national database.6, 7 The AUC for immediate mortality was 0.80 (95% confidence interval (CI) 0.71–0.90) in the smaller group. Analysis of the Finnvasc registry demonstrated that the predictive power for mortality was lower than previously suggested with an AUC of only 0.668 (p<0.0001). This was the only study published that firmly suggested that the GAS is inaccurate in this context. Since then Dutch and UK investigators have shown the GAS to be highly accurate in predicting mortality (AUC>0.80).8, 9 In addition the GAS was found to be a moderately accurate predictor of long-term mortality when applied to the open repair arm of the DREAM trial (AUC of 0.74).10 Some investigators have suggested that the outcome of symptomatic but non-ruptured AAA repair may also be predicted through GAS, where it could potentially be used to identify a high-risk group that may not benefit from immediate repair.11, 12

The simplicity of the GAS remains a great strength, making it easier to collate and use than most other scores. It has been validated successfully numerous times and predicts in-hospital mortality with acceptable accuracy in open AAA repair. However, when the GAS was compared with more recent models it performed relatively poorly despite an acceptable AUC of 0.749 (p=0.01).13

A drawback of the GAS is that it does not reliably identify individual high-risk patients due to a low-positive predictive value.14 It is also consistently inaccurate when used to predict morbidity (see Table 3).

Table 3. Summary of all studies
StudyModelO/EYearCountryNPeri-operative mortalityPeri-operative morbidityLong-term mortality
EagleEagleO1989US200
SamyGASO1994UK500Score proportionate to risk
SteyerburgLeidenO1995Netherlands246
SamyGASO1996UK320Score proportionate to risk
VanzettoVanzettoO1996France134 Perfusion scan predicts cardiac events
MidwinterPOSSUMO1999UK221Overestimates riskc
P-POSSUMPredictivec
PrytherchPOSSUM+P/VO2001UK444Predictivec
PrytherchV-POSSUM+PO2001UK1313PredictivecPredictivec
ShuhaiberPOSSUM+PO2002UK118PredictivecPredictivec
BiancariGASO2003Finland1911AUC<0.70AUC<0.70
BiancariGASO2003Finland403AUC 0.80AUC for: MI 0.72/MI-related death 0.78/Stroke 0.84Predictive
KeratiLeidenO2003Netherlands361c-statistic 0.72
NesiGASO2004Italy268AUC 0.75aAUC<0.70
EagleAUC<0.70aAUC<0.70
VanzettoAUC 0.79aAUC<0.70
LeidenAUC 0.78aAUC<0.70
Modified LeidenAUC 0.79aAUC<0.70
KertaiRevised Lee indexO2005Netherlands2310c-statistic 0.85
PrytherchVBHOMO2005UK831Predictive in electivec
LeoGASO2005Finland49AUC 0.79Predictive of MI (p=0.002)
AntonelloGASO2006Italy42AUC 0.87
HirzallaGASO2006Netherlands229AUC 0.84aAUC 0.66
BiancariGASO2006Finland5498AUC 0.7 Predictive
TangPOSSUM+PO2007UK452Lack of statistical fit
V-POSSUMPhysiology-only model predictivec
VBHOMO2007UK327c-statistic 0.80
TangGASO2007UK204AUC 0.84aAUC<0.70a
VBHOMAUC 0.82a – but lack of fitAUC<0.70a
E-PASSbAUC 0.91 (PRS only)aAUC 0.93a
BraunSVS co-morbidity scoreO+E2007Germany328No relationNo relation
FaizerGASE2007UK558AUC<0.50
Modified LeidenAUC 0.7
Modified SVS co-morbidity scoreAUC<0.70
BaasGASO+E2008Netherlands345AUC 0.79//0.87 Two-year AUC 0.74/0.78
BohmGASE2008UK266AUC<0.70AUC<0.70
Modified LeeAUC<0.70AUC<0.70
Lee indexAUC<0.70AUC<0.70
V-POSSUMAUC<0.70AUC<0.70
BarnesAustralian Audit ModelO+E2008Australia961Accurately predicts mortalityAccurately predicts multiple outcomes

Peri-operative mortality/morbidity=events within 30-days of surgery. AUC=area under ROC curve. O=open repair. E=EVAR.

a“In-hospital” mortality or morbidity described.

bThis population is described twice in the literature. The duplicate study was omitted.

cThe chi-squared test was used to test predictive power in these studies.

POSSUM prediction models 

The POSSUM score was created in 1991 primarily for use in surgical audit.15 It contains both physiological and operative components, and can only be used to predict mortality pre-operatively if the physiological component is used alone.16, 17 The dataset and the weighing of each variable are described well elsewhere.18 In 2001 P-POSSUM was constructed for general surgery, as POSSUM over-predicted death in low-risk patients.19 Despite this, a small single-centre study demonstrated good predictive ability of POSSUM and P-POSSUM in a mixed group of elective and emergency vascular cases.20

Modifications were subsequently suggested to improve predictive accuracy of POSSUM for vascular surgical procedures including AAA repair. V-POSSUM was developed by asking members of the Vascular Surgical Society for Great Britain and Ireland (VSSGBI) to suggest additional factors that may usefully add to the standard P-POSSUM dataset. The new model, V-POSSUM, was validated using data from the UK National Vascular Database (NVD), and it was found that the operative and physiology score predicted outcome by a comparison of predicted versus observed deaths (χ2=0.89, 4 df, p=0.926). The physiology score alone also predicted death (χ2=1.85, 4 df, p=0.765).21 The models failed to predict the results of emergency surgery in this instance. The extra data items suggested by the VSSGBI members did not contribute helpfully to the model. Despite P-POSSUM and V-POSSUM operative and physiology models predicting outcome well, the physiology-only P-POSSUM model was shown to be inaccurate in elective and emergency cases.21

Subsequent work demonstrated that the V-POSSUM physiology-only score was accurate in predicting death in emergency and elective surgery with a c-statistic of 0.877, but this time the predictions of POSSUM and P-POSSUM demonstrated significant lack of fit when compared to observed mortality rates.9 A new model, the Cambridge POSSUM score, has been proposed which has not yet been validated nor used as a physiology-only prediction tool.

The various POSSUM-based scores have produced variable results when undergoing validation as a pre-operative predictive score. Many available studies were concerned with the score as an audit tool. Use of physiology-only scores alone yielded inconsistent results when used to predict outcome of AAA repair. The POSSUM scores were the most complex scoring systems, and required a large number of variables. Missing variables are recorded as being normal, so the scores can potentially underestimate risk. Furthermore, some POSSUM variables were scored subjectively, a potential source of observer bias. Together these factors suggest that the POSSUM models are unlikely to form the basis a useful pre-operative risk prediction score at present.

Vascular Biochemical and Haematological Outcome Model (VBHOM) 

The VBHOM score was first described as an alternative to the POSSUM scores that were criticised for being unwieldy. VBHOM is a logistic regression model based on a minimal dataset derived from NVD haematology and biochemistry data from a single year. It was initially designed for audit.22 It was validated using a population of patients undergoing both elective and emergency AAA repair, and appeared to have a good predictive value. This conclusion was not supported by further study which suggested that although discriminative power was good (AUC=0.82: 95% CI 0.68–0.95; p=0.0001), it failed to show statistical “goodness-of-fit” indicating poor calibration.23 In response, a new VBHOM model was constructed using two years of NVD entries amounting to 2718 patients.24 External validation suggested a high degree of accuracy when it was used to predict death (c-statistic of 0.852) and the number of deaths predicted by the model closely agreed with observed mortality (χ2=8.40, 10 df, p=0.590; no evidence of lack of fit). This suggested that such a model based on the result of a single pre-operative blood sample may accurately predict the outcome of AAA surgery.

The VBHOM system requires a minimum dataset which is entirely objective, and is suitable for collection pre-operatively in both emergency and elective situations. The results of validation have not been consistent however, and the original version of the model has been shown to demonstrate poor calibration. The second version based on the UK NVD requires further validation.

Estimation of physiological ability and surgical stress (E-PASS) 

The E-PASS system was first used for general surgery, where it accurately predicted post-operative outcome.25 A comprehensive risk score (CRS) overall score was derived from a pre-operative risk score (PRS) and surgical stress score (SSS), encompassing six pre-operative and three peri-operative variables respectively. The CRS was shown to predict operative mortality very accurately with an AUC of 0.92, but could not be used pre-operatively because the SSS would not be known. However, the PRS alone had a positive predictive value similar to the overall CRS.13 E-PASS was more accurate than either the VBHOM or GAS when compared directly (AUC of 0.92 versus 0.76 and 0.84 respectively) in a 2007 study.23 The CRS also correlated with post-operative complications and length of hospital stay (p<0.001). E-PASS assumed surgical stress-induced SIRS, leading to multiple organ dysfunction. A drawback of this prediction method was that systemic inflammatory states potentially confounded the calculation of the PRS. Pre-existing systemic inflammatory response syndrome not measured by E-PASS may have resulted in worse outcome. As a result development of an emergency model that accounted for this was proposed. The E-PASS system has not yet been externally validated.

Leiden score and modifications 

The Leiden score used data derived from patients undergoing elective surgery for AAA at Leiden hospital between 1977 and 1988.26 odds ratios associated with pre-operative factors from a literature search were pooled and incorporated in the logistic regression model. It was unique in that it employed a corrective factor for institution-specific mortality rates. A modified Leiden score was created by rationalising the cardiovascular risk score, preventing patients from scoring several times with similar criteria. The threshold that defined renal impairment was reduced to be more consistent with other contemporary practice. The score was found to be a good predictor of mortality (AUC 0.788), but not morbidity.13 It was better at discriminating risk for those in the lower-risk bracket than high-risk patients. In 2005 the Leiden score was compared with a predictive model derived from the UK Small Aneurysm Trial data.27 It predicted death with moderate accuracy with an AUC of 0.72, and the latter model was shown to have poor predictive power and calibration.

The Lee or customised probability index (CPI) 

The Lee index predicts cardiac complications for non-cardiac surgery, and was modified in 2005 for predicting mortality in vascular surgery patients.28 This method included specific weighting based on the complexity of the operation. The score was derived and validated using data obtained from patients undergoing different vascular procedures including 837 elective and emergency repairs of AAA. The c-statistic was 0.83 in the validation set for all procedures. It was unique because it allowed for the positive prognostic effects of pre-operative statins and beta-blocker use. As well as being accurate it was simple to use, but requires further external validation in AAA repair.

Eagle and Vanzetto score 

The Eagle score used physiological factors in conjunction with myocardial thallium scanning in patients who had major vascular surgery to predict cardiac complications.29 This score appeared to predict the probability of cardiac events accurately. Other similar work by Vanzetto et al. produced comparable results.30 Both groups concluded that patients who may benefit from pre-operative coronary angiogram and revascularisation could be identified using these scores. Both the Eagle and Vanzetto scores have been externally validated, using only the clinical parameters to predict mortality after elective AAA. The Vanzetto score was more accurate than the Eagle score, which performed poorly.13

Models that predict risk associated with endovascular repair 

The suitability of the GAS for predicting mortality following EVAR has been assessed using data from the DREAM trial. Immediate post-operative death and two-year survival were predicted with an AUC of 0.87 and 0.78 respectively.10 Although the cut-off point denoting high-risk patients differed due to lower mortality in the EVAR group, the GAS was unexpectedly better at predicting risk of death for EVAR than for open repair in this instance. Similar claims were made after using the GAS to predict outcome of EVAR in the EUROSTAR registry population, despite a borderline AUC of 0.70 (95% CI=0.66–0.74; p<0.001). This is considered the minimum discriminative ability accepted in risk prediction models.14 Recently, two attempts were made to validate the “Co-morbidity Severity Score of the Society for Vascular Surgery and the American Association for Vascular Surgery” (SVS/AAVS co-morbidity score).31, 32 Both these studies revealed that this score was a poor predictor of outcome. In one of these studies the GAS achieved an AUC of <0.5, and the modified Leiden score was only weakly predictive. In 2008 several different models including GAS were compared using a single institute's EVAR data, comprising 266 cases. The GAS achieved a poor AUC of 0.678 (95% CI 0.48–0.97, p=0.046), with V-POSSUM displaying even less accuracy at 0.66 (95% CI 0.51–0.81, p=0.067). The customised probability index (CPI), and the modified form of this (m-CPI) were weak predictors of mortality. All of the scores failed to predict post-operative morbidity. This study concluded that none of these systems can be recommended for predicting results of EVAR at present.33

Morphology of an AAA influences the technical feasibility of EVAR, and dimensions of the proximal neck, extent of the aneurysmal disease distal to the aortic bifurcation and condition of the iliac arteries are known to contribute to graft-related complications and subsequent re-intervention.34 Therefore, some attempts have been made to assign weight to the various anatomical properties of aneurysms. A comparison of physiology and morphological scores identified aneurysm length, proximal neck length, neck angulation, neck calcification, intra-mural thrombus and distal landing zone as being significantly associated with outcome. The conclusion of this study was that those with large aneurysms and poor physiological risk profiles were generally less technically suitable for EVAR.35

The Australian National Audit of vascular surgery was recently used to describe factors that influenced long-term survival and graft-related outcomes.36 Age, ASA, gender and creatinine were found to be associated with risk. Maximum diameter, neck angle, infra-renal neck width and infra-renal neck diameter of an AAA were similarly related to adverse outcomes. As the audit was derived from a limited dataset, more complex anatomical features such as iliac tortuosity or presence of thrombus in the lumen of the aneurysm could not be analysed for significance. However, this model predicted a large number of outcomes and the computer interface is very user-friendly. External validation of this system is needed.

Back to Article Outline

Discussion 

A review of existing pre-operative risk predictor scores for open AAA repair revealed that none are entirely satisfactory. At present the GAS would appear to be the most useful when predicting risk of mortality after open AAA repair. It is easy to use and has been validated more consistently than any other method. Apart from a single study using the GAS, there is no evidence for the use of any existing scores to predict the risk associated with EVAR at present. Table 4 summarises the relative strengths and drawbacks of each score.

Table 4. Summary of relative usefulness of scores identifying major strengths and weaknesses
ModelStrengthsDrawbacks
Glasgow Aneurysm ScoreSimple. Consistently validated. Potentially predicts long-term outcome in open/EVARPoor in discriminating high-risk patients. Does not predict morbidity well. Not reliably validated for EVAR
POSSUM and P/V-POSSUMValidated audit tool. Physiology-only scores accurate in some studies for pre-operative predictionLarge dataset. Subjective elements. Most models require operative data. POSSUM overestimates risk
VBHOMMinimal dataset. Validated and highly accurateEarly model displayed poor calibration. Requires further validation
E-PASSHighly accurate for mortality and morbidity. Relatively small datasetCannot be used in SIRS. Requires further validation
Leiden/modified LeidenAdjusts for centre-specific mortality. Easy to use. Validated in EVAR (AUC only 0.70)Less predictive power in higher-risk patients
Eagle scoreIdentifies those at moderate risk of peri-operative MI requiring revascularisationNot useful if low or high risk of cardiac event. Poor predictor of all-cause peri-operative mortality
Vanzetto scoreIdentifies those requiring coronary intervention pre-operativelyModerately accurate when validated for all-cause peri-operative mortality
Modified Lee (or CPI)Allows for protective effect of medicationPerformed poorly when validated for EVAR. Requires further validation for open/EVAR
Australian Audit Risk Model for EVARComprehensive graft-related outcomes predicted using easy to use computer programRequires further validation
SVS co-morbidity scoreDesigned specifically for EVARDoes not appear to predict outcome of open or EVAR

SVS=Society of Vascular Surgeons; CPI=combined prognostic index.

A novel risk prediction system for EVAR and open repair based on physiological and anatomical variables, validated in different geographical areas and based on routinely obtained measurements, would assist the clinical practice of modern AAA repair. It could also be used for audit of case-mix, and a measure of utility such as the quality-adjusted-life-year (QALY) could be incorporated into aid analysis of cost-effectiveness.

Routine data collected for Hospital episode statistics (HES) have been used in England to construct a risk prediction model from 44,486 patient records who underwent open and endovascular AAA repair, as well as thoracic aortic surgery.37 This predicted the outcome of surgery with an area under the ROC curve of 0.74. Some of the variables contained in the HES could potentially be used in a model that would adjust for geographical location and social deprivation.

A risk score that determined an individual patient's predicted risk in specific hospitals would be useful in the consent process. The Leiden score assigned individual institutions a modifier that reflected local outcomes. During development of this system it was suggested that volume of cases undertaken and individual surgeon experience influenced centre-specific mortality to a greater degree than the characteristics of the patient population being served. This would result in complex operations performed in centres with a higher than average mortality attracting a prohibitive risk score. Some recent meta-analysis supported this by demonstrating a strong correlation between outcome and annual volume of AAA repair undertaken.38, 39 Other institutional structural risk factors have been incorporated into risk scores such as a “multi-level” model of the survival of patients operated on for ruptured AAA.40 This involved each unit being assigned a modifier based on a local “ICU effect”. Although centre-specific modifiers are an attractive idea they are also very difficult to calculate, and may not be acceptable to all surgeons.

Although EVAR has a lower mortality rate than open repair, the aim of future scoring systems should be to identify those patients who are at high risk of re-intervention or endoleak, as endograft durability is now of primary importance. A large cohort of patients should be used in order to identify high-risk subgroups and define important outcomes. Aneurysm morphology could be defined comprehensively in each case to identify all significant features. The importance of assessing anatomical suitability in a standard way is illustrated by a recent study that reports a higher complication rate if endografts are deployed in conflict with the manufacturer's recommended anatomical specifications.41 Recording and auditing the scores of those who suffer device-related complications as well as those deemed not suitable for surgery due to limitations of existing devices could direct development of new techniques and devices for management of complex aortic disease.

We propose the development of a new risk scoring system for AAA repair incorporating anatomical and physiological data to accurately predict a wide range of outcomes in the short and long term. This score will focus predominantly on EVAR, and the relationship between aneurysm morphology and endograft-related complications will be central to this. A major challenge will lie in making this score readily applicable to the routine practice of AAA surgery. Risk stratification scores will continue to be an important area of research in an era where the outcome and cost-effectiveness of surgical techniques are under more scrutiny than ever.

Back to Article Outline

Conflict of Interest 

All authors declare no competing interests.

Back to Article Outline

Appendix. Pre-operative risk score equations 


1.GAS=Age+(17 for shock)+(7 for myocardial disease)+(10 for cerebrovascular disease)+(14 for renal disease).

2.V-POSSUM physiology-only score=ln(R/1−R)=−6.0386+0.1539(physiology score). See www.riskprediction.org.uk for full description of score.

3.VBHOM=lne(R/1−R)=−2.257+0.1511(gender)+0.9940(mode of admission)+0.05923(age on admission)+0.001401(urea)−0.01303(sodium)−0.03585(potassium)−0.2278(Hb)+0.02059(WBC).

4.E-PASS – PRS component=0.0686+0.00435(age)+0.323(cardiac score)+0.205(pulmonary score)+0.153(diabetes score)+0.148(ASA)+0.666(performance status index). All scored 0–1 except ASA (0–5) and PSI (0–4).

5.Leiden score=Centre-specific score (−5 to +10)+age (years/2.5)+cardiac score (MI +3, CCF +8, ECG ischaemia +8)+respiratory impairment (+7)+gender (+4 if female). Centre-specific score not in modified, nor is ECG ischaemia.

6.Lee index=type of surgery (AAA +26)+IHD (+13)+CCF (+14)+stroke (+10)+hypertension (+7)+renal failure (+16)+respiratory disease (+7)+beta-blockers (−15)+statin (−10). Mortality risk in percent can then be calculated by a standard nomogram.

7.Eagle=0.77(age−10)+(angina +1)+(Q-waves +1.4)+(ventricular ectopics +1.2)+(diabetes+1)+(ECG changes to dipyramidole +1.3)+(evidence of ischaemia on thallium scan +2.3).

8.Vanzetto=+1 for each of; age>77, previous MI, angina, CCF, DM, hypertension, Q-waves on ECG and ST-segment changes on ECG. Those at moderate risk (2–4 risk factors) were thought to benefit most from thallium scan to assess need for coronary revascularisation.


10.SVS/AAVS co-morbidity score=cardiac score (0–3)+pulmonary score (0–3)+renal status (0–3)+hypertension (0–3)+age (0=<55, 1=55–69, 2=70–79, 3=>80). Score fully described elsewhere.32

Back to Article Outline

References 

  1. Ernst C. Abdominal aortic aneurysm. N Engl J Med. 1993;328:1167–1172
  2. Hoornweg LL, Storm-Versloot MN, Ubbink DT, Koelemay MJ, Legemate DA, Balm R. Meta analysis on mortality of ruptured abdominal aortic aneurysms. Eur J Vasc Endovasc Surg. 2008;35(5):558–570
  3. Wanhainen A, Bylund N, Bjorck M. Outcome after abdominal aortic aneurysm repair in Sweden 1994–2005. Br J Surg. 2008;95(5):564–570
  4. Samy AK, Murray G, MacBain G. Glasgow aneurysm score. Cardiovasc Surg. 1994;2(1):41–44
  5. Samy AK, Murray G, MacBain G. Prospective evaluation of the Glasgow Aneurysm Score. J R Coll Surg Edinb. 1996;41(2):105–107
  6. Biancari F, Leo E, Ylonen K, Vaarala MH, Rainio P, Juvonen T. Value of the Glasgow Aneurysm Score in predicting the immediate and long-term outcome after elective open repair of infrarenal abdominal aortic aneurysm. Br J Surg. 2003;90(7):838–844
  7. Biancari F, Heikkinen M, Lepantalo M, Salenius JP. Glasgow Aneurysm Score in patients undergoing elective open repair of abdominal aortic aneurysm: a Finnvasc study. Eur J Vasc Endovasc Surg. 2003;26(6):612–617
  8. Hirzalla O, Emous M, Ubbink DT, Legemate D. External validation of the Glasgow Aneurysm Score to predict outcome in elective open abdominal aortic aneurysm repair. J Vasc Surg. 2006;44(4):712–716
  9. Tang TY, Walsh SR, Prytherch DR, Wijewardena C, Gaunt ME, Varty K, et al. POSSUM models in open abdominal aortic aneurysm surgery. Eur J Vasc Endovasc Surg. 2007;34(5):499–504
  10. Baas AF, Janssen KJ, Prinssen M, Buskens E, Blankensteijn JD. The Glasgow Aneurysm Score as a tool to predict 30-day and 2-year mortality in the patients from the Dutch Randomized Endovascular Aneurysm Management trial. J Vasc Surg. 2008;47(2):277–281
  11. Antonello M, Lepidi S, Kechagias A, Frigatti P, Tripepi A, Biancari F, et al. Glasgow aneurysm score predicts the outcome after emergency open repair of symptomatic, unruptured abdominal aortic aneurysms. Eur J Vasc Endovasc Surg. 2007;33(3):272–276
  12. Leo E, Biancari F, Kechagias A, Ylonen K, Rainio P, Romsi P, et al. Outcome after emergency repair of symptomatic, unruptured abdominal aortic aneurysm: results in 42 patients and review of the literature. Scand Cardiovasc J. 2005;39(1–2):91–95
  13. Nesi F, Leo E, Biancari F, Bartolucci R, Rainio P, Satta J, et al. Preoperative risk stratification in patients undergoing elective infrarenal aortic aneurysm surgery: evaluation of five risk scoring methods. Eur J Vasc Endovasc Surg. 2004;28(1):52–58
  14. Biancari F, Hobo R, Juvonen T. Glasgow Aneurysm Score predicts survival after endovascular stenting of abdominal aortic aneurysm in patients from the EUROSTAR registry. Br J Surg. 2006;93(2):191–194
  15. Copeland G, Jones D, Walters M. POSSUM: a scoring system for surgical audit. Br J Surg. 1991;78(3):355–360
  16. Whiteley MS, Prytherch DR, Higgins B, Weaver PC, Prout WG. An evaluation of the POSSUM surgical scoring system. Br J Surg. 1996;83(6):812–815
  17. Midwinter M, Ashley S. Risk stratification in vascular patients using modified POSSUM scoring system. Br J Surg. 1997;84:A568
  18. Smith J, Tekkis PP. POSSUM predictor equation. Available from: <www.riskprediction.org.uk>2008;
  19. Prytherch DR, Ridler BM, Beard JD, Earnshaw JJ. A model for national outcome audit in vascular surgery. Eur J Vasc Endovasc Surg. 2001;21(6):477–483
  20. Shuhaiber JH, Hankins M, Robless P, Whitehead SM. Comparison of POSSUM with P-POSSUM for prediction of mortality in infrarenal abdominal aortic aneurysm repair. Ann Vasc Surg. 2002;16(6):736–741
  21. Prytherch DR, Sutton GL, Boyle JR. Portsmouth POSSUM models for abdominal aortic aneurysm surgery. Br J Surg. 2001;88(7):958–963
  22. Prytherch DR, Ridler BM, Ashley S. Risk-adjusted predictive models of mortality after index arterial operations using a minimal data set. Br J Surg. 2005;92(6):714–718
  23. Tang TY, Walsh SR, Fanshawe TR, Seppi V, Sadat U, Hayes PD, et al. Comparison of risk-scoring methods in predicting the immediate outcome after elective open abdominal aortic aneurysm surgery. Eur J Vasc Endovasc Surg. 2007;34(5):505–513
  24. Tang T, Walsh SR, Prytherch DR, Lees T, Varty K, Boyle JR. VBHOM, a data economic model for predicting the outcome after open abdominal aortic aneurysm surgery. Br J Surg. 2007;94(6):717–721
  25. Tang T, Walsh SR, Fanshawe TR, Gillard JH, Sadat U, Varty K, et al. Estimation of physiologic ability and surgical stress (E-PASS) as a predictor of immediate outcome after elective abdominal aortic aneurysm surgery. Am J Surg. 2007;194(2):176–182
  26. Steyerberg EW, Kievit J, de Mol Van Otterloo JC, van Bockel JH, Eijkemans MJ, Habbema JD. Perioperative mortality of elective abdominal aortic aneurysm surgery. A clinical prediction rule based on literature and individual patient data. Arch Intern Med. 1995;155(18):1998–2004
  27. Kertai MD, Steyerberg EW, Boersma E, Bax JJ, Vergouwe Y, van UH, et al. Validation of two risk models for perioperative mortality in patients undergoing elective abdominal aortic aneurysm surgery. Vasc Endovascular Surg. 2003;37(1):13–21
  28. Kertai MD, Boersma E, Klein J, van SM, Schouten O, van UH, et al. Optimizing the prediction of perioperative mortality in vascular surgery by using a customized probability model. Arch Intern Med. 2005;165(8):898–904
  29. Eagle KA, Coley CM, Newell JB, Brewster DC, Darling RC, Strauss HW, et al. Combining clinical and thallium data optimizes preoperative assessment of cardiac risk before major vascular surgery. Ann Intern Med. 1989;110(11):859–866
  30. Vanzetto G, Machecourt J, Blendea D, Fagret D, Borrel E, Magne JL, et al. Additive value of thallium single-photon emission computed tomography myocardial imaging for prediction of perioperative events in clinically selected high cardiac risk patients having abdominal aortic surgery. Am J Cardiol. 1996;77(2):143–148
  31. Braun K, Brunkwall J, Gawenda M. Risikoeinschätzung in der Aortenchirurgie – Evaluation des SVS/AAVS Comorbidity Severity Score. Zentralbl Chir. 2007;132(6):477–484
  32. Faizer R, DeRose G, Lawlor DK, Harris KA, Forbes TL. Objective scoring systems of medical risk: a clinical tool for selecting patients for open or endovascular abdominal aortic aneurysm repair. J Vasc Surg. 2007;45(6):1102–1108
  33. Bohm N, Wales L, Dunckley M, Morgan R, Loftus I, Thompson M. Objective risk-scoring systems for repair of abdominal aortic aneurysms: applicability in endovascular repair?. Eur J Vasc Endovasc Surg. 2008;36:172–177
  34. Stanley BM, Semmens JB, Mai Q, Goodman MA, Hartley DE, Wilkinson C, et al. Evaluation of patient selection guidelines for endoluminal AAA repair with the Zenith Stent-Graft: the Australasian experience. J Endovasc Ther. 2001;8(5):457–464
  35. Butcher W, Darke SG. A comparison of physiology scores and morphology in a group of patients evaluated for endovascular repair of infrarenal aneurysms. Int Angiol. 2004;23(1):66–71
  36. Barnes M, Boult M, Maddern G, Fitridge R. A model to predict outcomes for endovascular aneurysm repair using preoperative variables. Eur J Vasc Endovasc Surg. 2008;35:571–579
  37. Aylin P, Bottle A, Majeed A. Use of administrative data or clinical databases as predictors of risk of death in hospital: comparison of models. BMJ. 2007;334(7602):1044
  38. Young EL, Holt PJ, Poloniecki JD, Loftus IM, Thompson MM. Meta-analysis and systematic review of the relationship between surgeon annual caseload and mortality for elective open abdominal aortic aneurysm repairs. J Vasc Surg. 2007;46(6):1287–1294
  39. Holt PJ, Poloniecki JD, Gerrard D, Loftus IM, Thompson MM. Meta-analysis and systematic review of the relationship between volume and outcome in abdominal aortic aneurysm surgery. Br J Surg. 2007;94(4):395–403
  40. Hadjianastassiou VG, Franco L, Jerez JM, Evangelou IE, Goldhill DR, Tekkis PP, et al. Informed prognosis [corrected] after abdominal aortic aneurysm repair using predictive modeling techniques [corrected]. J Vasc Surg. 2006;43(3):467–473
  41. Abbruzzese TA, Kwolek CJ, Brewster DC, Chung TK, Kang J, Conrad MF, et al. Outcomes following endovascular abdominal aortic aneurysm repair (EVAR): an anatomic and device-specific analysis. J Vasc Surg. 2008;48(1):19–28

PII: S1078-5884(08)00463-2

doi:10.1016/j.ejvs.2008.08.016

European Journal of Vascular & Endovascular Surgery
Volume 36, Issue 6 , Pages 637-645, December 2008

Article Tools