If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corresponding author. G.J. Bryce, Department of General and Vascular Surgery, Western Infirmary, Dumbarton Road, Glasgow G11 6NT, UK. Tel.: +44 (0) 7973482206.
Department of Vascular Surgery, Western infirmary, Dumbarton Road, Glasgow G11 6NT, UKInstitute of Cardiovascular & Medical Sciences, BHF Glasgow Cardiovascular Research Centre, University of Glasgow, 126 University Place, Glasgow G12 8TA, UK
Department of Vascular Surgery, Western infirmary, Dumbarton Road, Glasgow G11 6NT, UKInstitute of Cardiovascular & Medical Sciences, BHF Glasgow Cardiovascular Research Centre, University of Glasgow, 126 University Place, Glasgow G12 8TA, UK
Institute of Cardiovascular & Medical Sciences, BHF Glasgow Cardiovascular Research Centre, University of Glasgow, 126 University Place, Glasgow G12 8TA, UK
Risk indices help quantify the risk of cardiovascular events and death prior to making decisions about prophylactic AAA repair. This paper aims to study the predictive capabilities of 5 validated indices.
Design and methods
A prospective observational multi-centre cohort study from August 2005 to September 2007 in Glasgow recruited 106 consecutive patients undergoing elective open AAA repair. The Glasgow Aneurysm Score (GAS), Vascular physiology only Physiological and Operative Severity Score for enUmeration of Mortality (V(p)-POSSUM), Vascular Biochemical and Haematological Outcome Model (VBHOM), Revised Cardiac Risk Index (RCRI) and Preoperative Risk Score of the Estimation of Physiological Ability and Surgical Stress Score (PRS of E-PASS) were calculated. Indices were compared using receiver operating characteristic (ROC) analysis and area under the curve (AUC) estimates. End points were all-cause mortality, Major Adverse Cardiac Events (MACE) and cardiac death.
Results
GAS, VBHOM and RCRI did not predict outcome. V(p)-POSSUM predicted MACE (AUC = 0.681), cardiac death (AUC = 0.762) and all-cause mortality (AUC = 0.780), as did E-PASS (AUC = 0.682, 0.821, 0.703 for MACE, cardiac death and all-cause mortality respectively).
Conclusion
Whilst V(p)-POSSUM and E-PASS predicted outcome, the less complex RCRI and GAS performed poorly which questions the utility of decision making based on these surgical risk indices.
Despite numerous attempts at devising an accurate preoperative risk score to help identify high-risk patients undergoing open elective abdominal aortic aneurysm (AAA) repair no ideal method has been found. This paper highlights once again that there is no model widely accepted, or repeatedly validated, in predicting mortality or cardiac events following AAA repair. In addition, the repeated difficulties in risk score prediction, including geographical variation and their development based on small sample sizes, raise the question of whether an entirely different form of stratifier should be sought.
Introduction
Prophylactic surgical repair of an abdominal aortic aneurysm (AAA) is associated with a mortality of around 5–6% in the UK.
The UK small aneurysm trial comparing surgical repair with observation of a small AAA recruited 1090 men and women aged 60–76 years with infrarenal AAAs measuring 4.0–5.5 cm whom were considered fit for surgery. Prospective follow-up of these patients randomised to either early surgery or routine surveillance demonstrated a greater risk of death from surgery, primarily cardiac in origin, than in the observation group for the first three years of follow-up.
Survival was subsequently worse in the surveillance group so that survival curves crossed at three years, and although this never reached statistical significance the importance of operative mortality and patient selection has been underlined. Greater discrimination of those at risk of perioperative death with poorer short-term survival would improve management.
Several surgical risk indices have been developed to help define the risk of perioperative death for a given patient. The Eagle
Additive value of thallium single-photon emission computed tomography myocardial imaging for prediction of perioperative events in clinically selected high cardiac risk patient having abdominal aortic surgery.
scores, which use physiological factors in conjunction with electrocardiogram (ECG) analysis and myocardial thallium scanning, have been shown to be useful in predicting cardiac events in major vascular surgery. However, despite further external validation in predicting mortality after elective open AAA repair,
are unique in that they employ a corrective factor for institution-specific mortality rates. However the Leiden score has not consistently shown satisfactory predictive capabilities. When applied to UK small aneurysm trial data it predicted death with only moderate accuracy (AUC 0.72) and was shown to have poor predictive power and calibration.
The Estimation of Physiological Ability and Surgical Stress (E-PASS) was found to accurately predict postoperative outcome in open elective AAA surgery, even with the preoperative component used alone.
The aim of this study is to compare five validated preoperative risk stratification scoring systems for patients undergoing open elective AAA repair, and to assess their ability to predict all-cause mortality and major adverse cardiac events (MACE) in the perioperative period.
Methods
Study population
A prospective, observational, multi-centre cohort study was performed involving the 3 major vascular units within Glasgow (Gartnavel General Hospital, Glasgow Royal Infirmary and the Southern General Hospital). A cohort of consecutive patients admitted for elective open AAA repair was identified between August 2005 and September 2007. All patients were approached the evening prior to surgery and informed consent obtained. Patients were excluded if they were receiving endovascular aortic aneurysm repair, required unplanned surgery undertaken within less than 24 h of admission, if they were unable to give informed consent, or if they had evidence of preoperative cardiac ischaemia measured by preoperative ECG and cardiac troponin I (cTnI).
Patients and preoperative assessment
Demographic data and factors relating to the cardiac risk of surgery were gathered for each patient by history and case-note review. Collection of patient data, including preoperative blood tests, ensured information was gathered to allow calculation of the 5 scoring systems named below. Detailed definitions are available at the referenced publications.
Physiological parameters including age, presence and severity of both cardiac and respiratory disease, ECG changes, systolic blood pressure, pulse rate, haemoglobin, white cell count, urea, sodium, potassium and Glasgow coma scale were collected for online calculation of the V(p)-POSSUM score at http://www.riskprediction.org.uk/vasc-index.php. The score was calculated using the following equation:
Vascular biochemical and haematological outcome model
The results of preoperative plasma urea, sodium, potassium, haemoglobin, white cell count with the addition of age or male gender were applied to the following equation:
R is the risk of death. Sex takes the value 1 for male and 0 for female, and mode of admission takes the value 0 for elective and 1 for non-elective admissions.
The RCRI was calculated by scoring one for each of; high-risk surgery, ischaemic heart disease, congestive heart failure, cerebrovascular disease, the use of insulin therapy for diabetes mellitus or a preoperative creatinine level > 2 mg/dl (>176 μmol/l).
Preoperative risk score of the estimation of physiological ability and surgical stress score
The PRS component of the E-PASS score was calculated based on the following equation:
Postoperative follow-up
Postoperative screening for cardiac events was performed by serial ECGs and cTnI measurement on the morning of the 2nd and 5th postoperative days. If a patient was due to be discharged prior to the fifth postoperative day, tests were conducted on the morning of discharge instead of day 5. Additional postoperative investigations were conducted at the discretion of the treating surgeon. Patients were reviewed at an outpatient clinic 6–8 weeks post-procedure.
End points
End points for the study were MACE (non-fatal myocardial infarction and cardiac death), cardiac death and death from any cause. The definition of a non-fatal myocardial infarction used was that of the Joint European Society of Cardiology/American College of Cardiology Committee.
Myocardial infarction redefined – a consensus document of ‘The Joint European Society of Cardiology/American College of Cardiology Committee’ for the redefinition of myocardial infarction.
Cardiac death was defined as death secondary to myocardial infarction, cardiogenic shock or intractable dysrhythmia, and was determined by a review of postoperative data by two cardiologists. If a patient had died of a non-cardiac cause the date and cause of death were noted. If no cause of death was noted then a death certificate review was performed.
Statistical analysis
Statistical analysis was performed using the SPSS® statistical software package (SPSS, Chicago, Illinois, USA). Normal distributed data were reported as mean ± standard deviation. Non-parametric data was reported as median and interquartile range. Mann–Whitney U test was used to test the differences between non-parametric continuous variables in different subgroups. A 2-sample t-test was used to compare continuous variables in normally distributed data. Receiver operating characteristics were plotted and the area under the curve (AUC) estimated. A p-value <0.05 was considered statistically significant.
Ethical approval
Local research and development, and central ethics committee approval was obtained for the study. All patients were provided with an information sheet and had signed a study consent form.
Results
Patient characteristics and preoperative blood testing
During the study period 111 patients met the inclusion criteria and were invited to participate. Two patients declined and a further 3 were found to have elevated preoperative cTnI and were not included, resulting in 95% inclusion. None of these five patients developed a cardiac complication or died following their procedure. The median age of the remaining 106 patients was 73 (66–77) years. The male to female ratio showed a male predominance of 5:1. The majority of patients had smoked at some point in their life (85%) of whom 35 (33%) currently smoked. One fifth of patients had suffered a previous MI with similar numbers having suffered from cerebrovascular disease. Other patient characteristics are detailed in Table 1. The majority of laboratory variables were within normal reference ranges for the local population [Table 1].
Table 1Preoperative patient characteristics and laboratory results, and their distribution according to outcome.
Complete datasets required for the calculation of all 5 preoperative risk indices were collected in all 106 patients. Of these 16 (15%) suffered MACE within 30 days, of whom 5 patients suffered a cardiac death and 11 had a non-fatal MI. The 30-day all-cause postoperative mortality was 8.5% (9 deaths), of which 7 deaths occurred prior to hospital discharge (6.6% in-hospital mortality).
Platelet levels and prothrombin times were significantly higher in those that suffered a cardiac death (p = 0.019 and p = 0.008 respectively) on univariate analysis. Serum urea was significantly higher in patients who died from all-cause mortality (p = 0.037). Urea levels were also higher in those who suffered cardiac death, although this was not significant (p = 0.050). Haemoglobin levels were lower in those that died of all-cause mortality, although again this was not significant (p = 0.058).
Glasgow aneurysm score, vascular biochemical and haematological outcome model and Lee’s revised cardiac risk index
The median GAS, VBHOM and RCRI scores were not significantly different in those patients who suffered MACE, cardiac death, or in those who died of all-cause mortality [Table 2]. ROC analysis revealed low accuracy in predicting MACE, cardiac death and all-cause mortality for all 3 scores. Sensitivities and specificities were poor and all 3 scores appeared to perform poorly in all outcome measures [Table 3].
Table 2Preoperative risk scores compared to patient outcome.
All subjects
MACE
Cardiac death
All-cause mortality
GAS
77 (72–84)
Yes
79 (71–84)
77 (70–85)
84 (74–87)
No
71 (72–84) p = 0.846
71 (72–84) p = 0.911
77 (71–83) p = 0.227
V(p)-POSSUM
29 (27–34)
Yes
33 (28–37)
34 (32–37)
34 (29–37)
No
29 (26–32) p = 0.028
29 (26–33) p = 0.030
29 (26–33) p = 0.038
VBHOM
−2.66 (−3.07 to −2.23)
Yes
−2.53 (−2.96 to −2.08)
−2.57 (−2.78 to −1.85)
−2.33 (−2.62 to −2.21)
No
−2.67 (−3.12 to −2.26) p = 0.232
−2.66 (−3.08 to −2.23) p = 0.340
−2.72 (−3.10 to −2.30) p = 0.069
RCRI
2 (1–3)
Yes
2 (1–3)
3 (1–4)
2 (1–4)
No
2 (1–3) p = 0.735
2 (1–3) p = 0.448
2 (1–3) p = 0.415
PRS
0.46 (0.26–0.64)
Yes
0.63 (0.46–0.67)
0.65 (0.59–1.02)
0.65 (0.40–0.93)
No
0.42 (0.25–0.60) p = 0.019
0.44 (0.26–0.62) p = 0.017
0.44 (0.26–0.62) p = 0.050
MACE – major adverse cardiac event, GAS – Glasgow aneurysm score, V(p)-POSSUM – physiological component of vascular physiological and operative severity score for enumeration of mortality, VBHOM – vascular biochemical and haematological outcome model, RCRI – revised cardiac risk index, PRS – preoperative risk score of estimation of physiological ability and surgical stress.
All scores recorded as median (interquartile range).
Vascular (physiology only) – physiological and operative severity score for enumeration of mortality
There was a difference in median V(p)-POSSUM with higher scores in those who suffered MACE (p = 0.028), cardiac death (p = 0.030), and in those who died of all-cause mortality (p = 0.038) [Table 2]. ROC analysis revealed an AUC of 0.681 in predicting MACE with the best cut-off for V(p)-POSSUM of 20.5, an AUC of 0.762 in predicting cardiac death and an AUC of 0.780 for predicting all-cause mortality, with once again a cut-off of 20.5 for both outcome measures [Table 3].
Preoperative risk score of the estimation of physiological ability and surgical stress score
The median PRS was higher in those who suffered MACE (p = 0.019) and cardiac death (p = 0.017). The median PRS was higher in the all-cause mortality group than in the survivors, although this was just not significant (p = 0.050) [Table 2]. ROC analysis revealed an AUC of 0.682 in predicting MACE and 0.703 in predicting all-cause mortality, with a cut-off of 0.55 for both outcome measures. ROC analysis revealed an AUC of 0.821 for predicting cardiac death [Fig. 1] with a cut-off of 0.63 [Table 3].
Figure 1Receiver operating characteristics curve for PRS and cardiac death with an area under the curve of 0.821.
Tertile analysis was performed by dividing each index grouping into 3 equal classes based on even distribution with the aim of examining stepwise increase in MACE, cardiac death and mortality between groups. For GAS (p = 0.584, p = 0.891 and p = 0.264 respectively), VBHOM (p = 0.395, p = 0.174 and p = 0.070) and RCRI (p = 1.0, p = 0.368 and p = 0.867) there were no significant differences in the tertile rates of MACE, cardiac death and all-cause mortality.
For V(p)-POSSUM sequential increases were seen in rates for MACE and all-cause mortality with rates in MACE of 4%, 13% and 29%, and in all-cause mortality of 0%, 8% and 16% (p = 0.013 and p = 0.030 respectively). In cardiac death there was no significant rate change between tertiles with rates of 0%, 4% and 10% (p = 0.130) [Fig. 2a].
Figure 2Tertile analysis for individual risk scores. MACE – major adverse cardiac event, V(p)-POSSUM – physiological component of vascular physiological and operative severity score for enumeration of mortality, PRS – preoperative risk score of estimation of physiological ability and surgical stress. All analyses were performed using Mann–Whitney U test.
For the PRS component of E-PASS there were significant increases in rates for all outcome measures with rates of 6%, 13% and 27% in MACE, 0%, 3% and 12% in cardiac death, and rates of 6%, 3% and 18% in all-cause mortality (p = 0.038, p = 0.015 and p = 0.032 respectively) [Fig. 2b].
Discussion
Preoperative risk stratification is of importance in elective open AAA repair and, since opinion and experience alone are often the only tools of the vascular surgeon in selecting operative candidates, an optimal risk stratifier has been sought. A number of risk scoring systems derived from patients undergoing non-cardiac surgery have been devised, but most seem to lack accuracy.
Based on a review of the literature, the 5 most suitable preoperative risk scoring indices have been identified and prospectively evaluated with an emphasis on predicting MACE, cardiac death and all-cause mortality in the immediate postoperative period.
The two scoring systems that performed well were V(p)-POSSUM and the Preoperative Risk Score component of E-PASS. In this cohort, the physiology component of V(p)-POSSUM was a significant predictor of all outcome measures. POSSUM scoring has long been advocated as the most fitting scoring system in non-cardiac surgery.
This may be due to the inclusion of a relatively high number of important serum and physiological markers, providing a more detailed account of physiological reserve. However, POSSUM has been shown to over-predict death in low risk patients,
It is therefore expected that V(p)-POSSUM would perform well in predicting all-cause mortality in the elective open AAA patient, given the higher risk associated with this surgery type compared to other types of non-cardiac and vascular procedures. It was an unexpected finding that it would perform well in predicting MACE, although this may reflect its ability to predict morbidity in general. Whilst this index appears to perform well, there is overriding concern with the effect of area-wide variation, and whilst V(p)-POSSUM allows for variations in the hospital case mix it is not robust in different geographical locations.
Further, its original development as that of an audit tool including operative data continues to raise questions about its use in the preoperative setting when guiding clinical decision making.
The PRS component of the E-PASS score predicted MACE and cardiac death, and whilst it did not appear to predict all-cause mortality on univariate testing it did perform well on tertile analysis and in ROC curve analysis. The main concern with E-PASS is that the original scoring system requires operative data to calculate its overall score thus precluding its use for preoperative risk assessment. However a strong correlation between the PRS component and outcome has been shown which therefore permits its use in the preoperative setting.
It may be the very premise of this scoring system, where each patient’s physiological reserve is scored, that gives a more accurate outcome measure.
Of the other scoring systems only VBHOM showed utility in predicting all-cause mortality. However, this did not reach statistical significance and VBHOM performed poorly in predicting MACE or cardiac death. This confirms previous validations where VBHOM was shown to under-predict perioperative complications and over-predict mortality.
This may be explained by the fact that VBHOM was devised including patients who underwent emergency AAA repair.
The GAS and RCRI performed poorly in this prospective cohort. The GAS was constructed using a population of 500 patients undergoing both emergency and elective AAA repair in Glasgow over a 10 year period between 1980 and 1990.
Value of the Glasgow aneurysm score in predicting the immediate and long-term outcome after elective open repair of infrarenal abdominal aortic aneurysm.
The Glasgow aneurysm score as a tool to predict 30-day and 2-year mortality in the patients from the Dutch randomised endovascular aneurysm management trial.
The Glasgow aneurysm score as a tool to predict 30-day and 2-year mortality in the patients from the Dutch randomised endovascular aneurysm management trial.
The GAS also fails to allow for treatment or optimisation of a patient’s comorbid condition and scores for the presence of disease rather than the functional effects of the disease process. More surprising is that the score is a poor predictor in the very population from which it was derived, which may be in part explained by the increasing age and improving health states of the population.
The most unexpected finding is that the RCRI performed worst. As the only score of the 5 included that was specifically designed to evaluate cardiac risk it is unexpected that it not only failed to predict mortality but also failed to predict cardiac events. This lack of predictive value may be due to the high-risk population within which it was used, immediately scoring every patient 1 prior to further scoring. It is likely that this diminishes the sensitivity of the index, as has been suggested in the published literature, where concern has been expressed at the grouping of intraperitoneal and major vascular surgery together, and where open AAA repair has shown to have worse outcome than all other operation subgroups that are considered together in the RCRI.
This lack of subgrouping is taken into account in the ESC/ESA Guidelines for preoperative cardiac risk assessment and perioperative cardiac management in non-cardiac surgery.
This may improve the accuracy of the RCRI, which plays a key role in the algorithm proposed by Poldermans et al. The results of this study do however raise concern about the use of these guidelines in open elective AAA repair.
The principal limitation of this prospective cohort is the small sample size. Although MACE occurred with sufficient frequency to draw reasonable conclusions, all-cause mortality and in particular cardiac death were low in absolute numbers. A larger cohort would have been more conducive to risk estimation. In contrast, however, the end points were well-defined and most vascular surgical practices will deal with numbers similar to this cohort. This therefore questions the utility of scoring systems that require large numbers to be clinically relevant.
In conclusion, it appears both the PRS component of E-PASS and V(p)-POSSUM have advantages in predicting outcome following elective open AAA repair over other scoring systems. Whilst V(p)-POSSUM appears to be the best stratification score (and easily accessible on the internet) with regard to all-cause mortality, its inability to predict mortality rates in individual patients may make E-PASS a more favourable alternative, having been developed specifically as an aid to clinical decision making, although further validation is required.
Both the RCRI and GAS lacked predictive power. As the simplest forms of risk score these are possibly the most often used in the clinical environment, and the results are therefore disappointing for real life clinical practice. The increasingly aged population with greater use of medical therapy and synchronous rise in EVAR may mean that more up to date indices such as the customised probability model
will show greater predictive abilities, however these await widespread validation and acceptance in risk assessing for open elective AAA repair. Based on the results of this study it would be advisable not to base clinical decision making in aneurysm surgery on RCRI or GAS scoring. Further, the complexity of those scores that do show predictive abilities make their use in the clinical setting on an individual patient basis debatable, and it may be these scores should remain audit tools rather than tools on which to guide decision making.
Additive value of thallium single-photon emission computed tomography myocardial imaging for prediction of perioperative events in clinically selected high cardiac risk patient having abdominal aortic surgery.
Myocardial infarction redefined – a consensus document of ‘The Joint European Society of Cardiology/American College of Cardiology Committee’ for the redefinition of myocardial infarction.
Value of the Glasgow aneurysm score in predicting the immediate and long-term outcome after elective open repair of infrarenal abdominal aortic aneurysm.
The Glasgow aneurysm score as a tool to predict 30-day and 2-year mortality in the patients from the Dutch randomised endovascular aneurysm management trial.
As older patients with significant comorbidities are increasingly undergoing elective open or endovascular repair to prevent abdominal aortic aneurysm (AAA) rupture, accurate preoperative stratification of the operative risk is of major importance. In fact, AAA surgery is associated with variable postoperative mortality and morbidity rates, related to the invasive nature of surgery, the frequent existence of severe comorbidities, the experience of surgeons and anaesthesiologists caring for the patients, and the hospital procedural volume.
To submit a comment for a journal article, please use the space above and note the following:
We will review submitted comments as soon as possible, striving for within two business days.
This forum is intended for constructive dialogue. Comments that are commercial or promotional in nature, pertain to specific medical cases, are not relevant to the article for which they have been submitted, or are otherwise inappropriate will not be posted.
We require that commenters identify themselves with names and affiliations.
00233-X&id=gr1.jpg){kind=link}
00233-X&id=gr2.jpg){kind=link}