Summary
Background
The operative risk of cardiac surgery is ascertained preoperatively on the basis of scores validated in multinational studies. However, the value they add to a simple bedside clinical evaluation (CE) remains controversial.
Aims
To compare operative mortality (defined as death from all causes before the 31st postoperative day) predicted by CE with that predicted by additive and logistic EuroSCOREs, EuroSCORE II and Society of Thoracic Surgeons (STS), Ambler and age-creatinine-ejection fraction (ACEF) scores in patients undergoing aortic valve replacement (AVR) for severe aortic stenosis.
Methods
Overall, 314 consecutive patients were included who underwent AVR between October 2009 and November 2011 (22% with coronary artery bypass graft); mean age 73.4 ± 9.7 years (29% aged > 80 years). Based on CE, patients were divided into four predefined groups of increasing estimated mortality risk: I ≤ 3.9%; II 4–6.9%; III 7–9.9%; IV ≥ 10%. The positive and negative predictive values of the six scores and CE were compared.
Results
The observed overall operative mortality was 5.7%. The distribution of the four predicted mortality groups by each score was highly variable. The positive predictive value, calculated for the 64 patients classified at highest risk by CE (groups III or IV) or each score, was 17.2% for EuroSCORE II, 14.1% for CE and STS scores, 10.9% for additive and logistic EuroSCOREs, 10.6% for ACEF and 10.2% for Ambler. The positive predictive value of each score in the low-risk groups (I and II) ranged from 2.8% to 4.4%.
Conclusion
A simple bedside CE appears as reliable as the various established scores for predicting operative risk in patients undergoing surgical aortic valve replacement. The development and validation of more comprehensive risk stratification tools, including risk factors thus far neglected, seems warranted.
Résumé
Contexte
Le risque opératoire de la chirurgie cardiaque est évalué en préopératoire sur la base de scores internationaux validés. Néanmoins, leur valeur ajoutée par rapport à l’évaluation clinique simple reste controversée.
Objectifs
Cette étude a comparé la mortalité opératoire (définie comme le décès de toute cause avant le 31 e jour postopératoire) prédite par l’évaluation clinique, avec celle prédite par l’EuroSCORE additif et logistique, l’EuroSCORE II, le score de la Society of Thoracic Surgeons (STS), le score d’Ambler et le score ACEF (âge-créatinine-fraction d’éjection) chez des patients opérés d’un remplacement valvulaire aortique pour sténose aortique serrée.
Méthodes
Ont été inclus 314 patients consécutifs (73,4 ± 9,7 ans d’âge, dont 29 % de plus de 80 ans) ayant subi un remplacement valvulaire aortique entre octobre 2009 et novembre 2011, associé à un ou des pontages coronaires pour 22 % d’entre eux. En fonction de l’évaluation clinique, les patients étaient répartis en 4 groupes prédéfinis de risque opératoire croissant: I ≤ 3,9 % ; II 4–6,9 % ; III 7–9,9 % ; IV ≥ 10 %. Les valeurs prédictives positive et négative des 6 scores et de l’évaluation clinique ont été comparées.
Résultats
La mortalité opératoire observée était de 5,7 %. La répartition selon les 4 groupes de mortalité prédite était hautement variable d’un score à l’autre. La valeur prédictive positive calculée pour les 64 patients classés les plus à risque par l’EC (groupes III et IV) ou par chacun des scores, était de 17,2 % pour l’EuroSCORE II, 14,1 % pour l’évaluation clinique et le score STS, 10,9 % pour l’EuroSCORE additif et logistique, 10,6 % pour l’ACEF et 10,2 % pour l’Ambler. La valeur prédictive positive de chaque score dans le groupe bas risque (I et II) se situait entre 2,8 et 4,4 %.
Conclusion
Une évaluation clinique simple semble aussi fiable que les scores dans la prédiction du risque opératoire du remplacement valvulaire aortique. Le développement et la validation d’outils plus élaborés d’estimation du risque, incluant des facteurs jusque là négligés, semblent mérités.
Background
Aortic stenosis is the most prevalent and the most often operated on valvular heart disease in Europe and North America. The 5-year survival of patients who present with severe non-operated aortic stenosis is approximately 30%, while the current operative mortality associated with aortic valve replacement (AVR) is < 5%, although it is higher in the elderly and in patients presenting with major co-morbidity . Furthermore, transcatheter aortic valve implantation (TAVI) may now be an option for patients who, until recently, were considered inoperable. Therefore, the estimation of operative risk is at the centre of the decision-making process. According to the European professional practice guidelines “the risk assessment should mostly rely on the clinical judgement of the ‘heart team’ in addition to the combination of scores” . Nevertheless, to the best of our knowledge, the performance of a simple bedside clinical evaluation (CE) has never been compared with the scores. While their merits have been highlighted, the limitations of the existing scoring systems have also been emphasized, particularly the differences among the variables used and their respective weights – a source of major discordance among the various models . The aim of this first prospective study was to compare the predictive value of a standardized “bedside” preoperative CE, with that of a preoperative estimate based on the main risk scores, in patients who underwent valve replacement for severe aortic stenosis.
Methods
Between September 2009 and November 2011, consecutive patients who underwent preoperative evaluations of severe aortic stenosis in the Department of Cardiology of Rennes University Hospital were prospectively included in this analysis, before being referred to the Department of Cardiothoracic Surgery of the same hospital, to undergo mechanical or bioprosthetic AVR, with or without coronary artery bypass graft (CABG). The echocardiographical criteria used to diagnose severe aortic stenosis were: a valve area < 1 cm 2 ; a mean transvalvular gradient > 40 mmHg; or a jet velocity > 4 m/s, as defined by the 2008 guidelines of the American College of Cardiology/American Heart Association . Patients who underwent additional cardiovascular operative procedures other than CABG were excluded from the study. The study protocol was approved by our institutional review board and the patients granted their informed consent to participate.
The CE was based on a consensus reached among cardiologists who examined the patient and ordered the preoperative workup. It consisted of classifying patients on a four-grade scale of estimated mortality risk: grade I (0–3.9%), grade II (4–6.9%), grade III (7–9.9%), or grade IV (≥10%), based on the results of all diagnostic investigations, which usually comprised history, physical examination, echocardiogram, coronary angiograms, respiratory functional exploration with spirometry, complete blood count, haemostasis and renal function tests with creatinine clearance and status of peripheral arterial vessels (namely ultrasonography of the carotid and vertebral arteries). Final grading reflected the opinion of at least three to four physicians. Fifteen experienced cardiologists in two teams (age range 30–61 years) participated in the CE process and grading. The reproducibility of CE was tested with six surgeons and eight anaesthesiologists. All physicians who participated in CE were unaware of the results of calculated scores. The interobserver reproducibility of CE grading was tested in a sample of 84 patients who underwent duplicate CE by the surgeon and another sample of 68 patients who underwent duplicate CE by the anaesthesiologist, both unaware of the results of the first CE by the cardiologists. Furthermore, we calculated independently and verified systematically six operative risk scores: additive EuroSCORE (ES-add), logistic EuroSCORE (ES-log), EuroSCORE II (ES II), Society of Thoracic Surgeons (STS) score, Ambler score, and age-creatinine-ejection fraction (ACEF) score. ES II and ACEF were calculated retrospectively at the end of the study.
EuroSCORE, the most often used score, is based on 17 variables, with either a simple additive mode (ES-add) or a computer-based algorithm (ES-log). This score was developed in 1995, from a database of 14,000 patients. ES II, its updated version, was published in 2011 . The Society of Thoracic Surgeons (STS) score includes approximately 40 variables that are regularly updated . The Ambler score was designed for risk stratification in valve surgery by British investigators in 2005, based on 16,000 patients; it can be calculated rapidly on the basis of 13 variables . All of these scores are accessible via a website. The more recent ACEF score, based on age, serum creatinine concentration and left ventricular ejection fraction (LVEF), has been found, in some studies, to be superior to the other scores . The various components of the different scores are summarized in Table 1 . The Charlson-age co-morbidity index (a useful complement to the risk scores for non-cardiovascular co-morbidity, which is often used in geriatric medicine, oncology and non-cardiac surgery), was also calculated, although it was not included in all between-scores comparisons.
| STS score | ES-add/log | ES II | Ambler | ACEF | |
|---|---|---|---|---|---|
| Patient characteristics | |||||
| Age | + | + | + | + | + |
| Sex | + | + | + | + | |
| Height | + | ||||
| Weight | + | ||||
| Body mass index | + | ||||
| Ethnicity | + | ||||
| Co-morbid conditions | |||||
| Diabetes | + | + (I) | |||
| Hypertension | + | + | |||
| Chronic lung/pulmonary disease | + | + | + | ||
| Extracardiac arteriopathy | + | + | + | ||
| Peripheral vascular disease | + | ||||
| Neurological dysfunction | + | + | |||
| Cerebrovascular accident | + | ||||
| Serum creatinine | + | + | + | + | + |
| Dialysis-dependent renal failure | + | + | |||
| Immunosuppressive therapy | + | ||||
| Poor mobility | + | ||||
| Cardiac history | |||||
| NYHA classification | + | + | |||
| Unstable angina | + | + | |||
| CCS class 4 angina | + | ||||
| Recent myocardial infarction | + | + | + | ||
| Arrhythmias | + | + | |||
| Previous cardiac surgery | + | + | + | ||
| Previous CABG | + | ||||
| Previous valvular surgery | + | ||||
| Previous PCI | + | ||||
| Active endocarditis | + | + | + | ||
| Number of diseased coronary vessels | + | ||||
| Type and severity of valvular disease | (stenosis/regurgitation, aortic/mitral) | + | |||
| Haemodynamic state | |||||
| Pulmonary hypertension | + | + | |||
| Ejection fraction | + | + | + | + | + |
| Critical preoperative state | + | + | |||
| Cardiogenic shock | + | ||||
| Resuscitation | + | ||||
| Inotropic agents | + | ||||
| Intra-aortic balloon pump | + | ||||
| Procedure | + | + | + | ||
| Emergency | + | + | + | ||
| Surgery on thoracic aorta | + | + | + | ||
| Aortic valve surgery | + | + | |||
| Mitral valve surgery | + | + | |||
| Aortic and mitral valve surgery | + | + | |||
| Tricuspid surgery | + | ||||
| Surgery for congenital heart disease | + | ||||
| Rhythmological surgery | + | ||||
| Concomitant CABG | + | + | |||
| Concomitant tricuspid valve surgery | + | + | |||
| Postinfarct septal rupture | + | ||||
Operative mortality was defined as death from all causes occurring before the 31st postoperative day, based on a review of records, postal correspondence or telephone contact with the patient, the cardiologist or the family physician. No patient was lost to follow-up.
CE grading was first compared with each score with respect to the four previously described categories of estimated risk. A second comparison was done for the same number of patients at highest risk ( n = 64) according to each scoring method or CE grading, defining a high-risk group depending on the score or method chosen. The other patients defined the low-risk group.
Statistical analyses
Continuous variables, expressed as means ± standard deviations, were compared using Student’s t -test. Qualitative variables, expressed as counts and percentages, were compared using the chi 2 test or Fisher’s exact test, as appropriate. The kappa test was used to examine the agreement between the scores and CE; a value of 1 indicated 100% agreement, while a value of 0 indicated agreement equivalent to chance. Receiver operating characteristic analysis, producing an area under the curve with 95% confidence intervals (CIs), was used for discrimination of the predictive scores of 30-day mortality. Risks (and 95% CIs) were calculated to compare risk groups. A P value < 0.05 was considered significant. The analyses were performed with the SAS software, version 9.2 (SAS Institute, Cary, NC, USA).
Results
Patients
A total of 314 consecutive patients (mean age 73.4 ± 9.7 years; range 22–90 years) were included, of whom 91 (29%) were aged > 80 years, 129 (41%) were women and 19 (6%) were asymptomatic. The baseline characteristics of the sample population are shown in Table 2 . The mean aortic valve area was 0.72 ± 0.16 cm 2 (range 0.35–1.2 cm 2 ), the mean transvalvular gradient was 54 ± 29 mmHg (range 18–100 mmHg) and the mean peak transaortic flow was 4.5 ± 1.4 m/s (range 2.6–6.1 m/s). The mean LVEF was 60.7 ± 10% (range 20–78%) and 18 patients had an LVEF ≤ 40%. A bioprosthesis was implanted in 260 (83%) patients, a mechanical valve was implanted in 54 (17%) patients and at least one CABG was performed in 69 (22%) patients. Operative mortality was 5.7% ( n = 18) and the mean duration of index hospitalization was 12 ± 6 days. Mortality was 10.1% (7/69) in patients undergoing AVR plus CABG and 4.5% (11/245) in patients undergoing AVR alone.
