1. Anaesthesia 2012, 67, 389–395 doi:10.1111/j.1365-2044.2011.07020.x
Original Article
What is the best pre-operative risk stratification tool for major
adverse cardiac events following elective vascular surgery? A
prospective observational cohort study evaluating pre-operative
myocardial ischaemia monitoring and biomarker analysis
B. M. Biccard,1 P. Naidoo2 and K. de Vasconcellos1
1 Consultant, Perioperative Research Unit, Department of Anaesthesia, 2 Specialist, National Health Laboratory Services,
Department of Chemical Pathology, Nelson R Mandela School of Medicine, University of KwaZulu-Natal, Durban,
KwaZulu-Natal, South Africa
Summary
Although brain natriuretic peptide has been shown to be superior to the revised cardiac risk index for risk stratification
of vascular surgical patients, it remains unknown whether it is superior to alternative dynamic risk predictors, such as
other pre-operative biomarkers (C-reactive protein and troponins) or myocardial ischaemia monitoring. The aim of this
prospective observational study was to determine the relative clinical utility of these risk predictors for the prediction of
postoperative cardiac events in elective vascular surgical patients. Only pre-operative troponin elevation (OR 56.8, 95%
CI 6.5–496.0, p < 0.001) and brain natriuretic peptide above the optimal discriminatory point (OR 6.0, 95% CI 2.7–12.9,
p < 0.001) were independently associated with cardiac events. Both brain natriuretic peptide and troponin risk
stratification significantly improved overall net reclassification (74.6% (95% CI 51.6%–97.5%) and 38.5% (95% CI 22.4–
54.6%, respectively)); however, troponin stratification decreased the correct classification of patients with cardiac
complications ()59%, p < 0.001). Pre-operative brain natriuretic peptide evaluation was the only clinically useful
predictor of postoperative cardiac complications.
. ..............................................................................................................................................................
Correspondence to: Dr B. Biccard
Email: biccardb@ukzn.ac.za
Accepted: 19 November 2011
It is estimated that nearly a million patients each year risk-stratify patients in a clinically useful manner, only
worldwide sustain major cardiac complications such as the revised cardiac risk index (RCRI) [2] has been
cardiac death, myocardial infarction and cardiac arrest incorporated into both American and European guide-
following non-cardiac surgery [1]. A number of pre- lines for pre-operative non-cardiac risk assessment [5, 6].
operative strategies have been used to identify at risk However, with regards to risk stratification for
patients, including clinical risk scores [2], the detection vascular surgical patients, there has been significant
of myocardial ischaemia by ambulatory Holter progress in determining appropriate stratification be-
monitoring [3], and biomarker analysis [4]. Although yond the RCRI, by incorporating the pre-operative
individually, all of these approaches have been used to brain natriuretic peptide (BNP) concentration. First, it
Anaesthesia ª 2012 The Association of Anaesthetists of Great Britain and Ireland 389
2. Anaesthesia 2012, 67, 389–395 Biccard et al. | Biomarkers and Holter monitoring for risk stratification
has been shown, using appropriate reclassification was added to the protocol, also measured in the 24 h
statistics [7, 8], that BNP has clinical utility for pre- before surgery [11].
operative risk stratification of vascular patients [9], even Starting in June 2008, a sub-cohort of patients
in the presence of RCRI risk stratification. Subse- underwent pre-operative Holter monitoring for myo-
quently, a collaborative group conducted an individual cardial ischaemia using a Schiller MT-200 ECG Holter
patient data meta-analysis, and based on the sample size monitor (Shiller AG, Baar, Switzerland). Monitoring
was able to determine that neither the RCRI score, nor started the day before surgery and was continued right
any of its components, was able to improve on a up to patients’ arrival in the operating theatre for
BNP-based pre-operative risk stratification for elective surgery. The number of patients recruited for this sub-
vascular surgical patients [10]. These studies have cohort was limited by the availability of the monitors.
established the superiority of BNP over the RCRI and The Holter data were only analysed (by KV) at the end
its individual component risk factors in pre-operative of the study, and the analysis was blinded as to outcome.
risk stratification. An a priori decision was taken to analyse the number of
However, it has not yet been shown whether BNP episodes of ST depression lasting > 10 min. We defined
will still retain its powerful predictive ability when ST depression as a deviation > 1 mV from baseline
compared with other dynamic risk predictors such as measured 60 ms after the J point. The end of the episode
pre-operative troponin levels and myocardial ischaemia of ST depression was defined as the return of the ST
monitoring. To address this question, we conducted a deviation to < 1 mV from baseline for 60 s. The ST
prospective observational study to determine the relative segments were inspected visually and ST deviations due
importance and clinical utility of the RCRI, pre-operative to artefact were not considered. Modified V5 and V2
myocardial ischaemia and pre-operative elevation of the leads were analysed. The mean heart rate, maximum
biomarkers C-reactive protein (CRP), BNP and tropo- heart rate and longest duration of time above
nins, in the prediction of postoperative major adverse 100 beats.min)1 were also analysed.
cardiac events within 30 days of elective vascular surgery. The peri-operative anaesthetic technique was at the
discretion of the attending anaesthetist. There was no
Methods study protocol for the management of an elevated
This study was conducted at Inkosi Albert Luthuli postoperative troponin, and management was deter-
Central Hospital, in KwaZulu-Natal, South Africa, with mined by the anaesthetic and surgical team on an
institutional ethics approval, and was registered with the individual patient basis. Attending clinicians were not
national administrative body (South African National blinded to the pre-operative biomarker results.
Clinical Trials Register). We recruited elective vascular The samples for BNP and Troponin I were collected
surgical patients between February 2008 and March in EDTA and serum separator tubes (Greiner Bio-One,
2011 who gave informed consent. Patients consented Frickenhausen, Germany), respectively. All samples
for: collection of clinical risk factors alone; collection of were centrifuged and analysed on receipt, using the
clinical risk factors with pre-operative biomarkers; or Advia CentaurÒ Xp (Siemens Healthcare, Malvern, PA,
collection of clinical risk factors and pre-operative USA), which involves chemiluminescent technology.
biomarkers, with ambulatory Holter monitoring. The CRP analysis was performed on serum samples,
All patients’ characteristics and cardiac clinical risk using the latex enhanced immunoturbidimetric method.
predictors were collected as per the definition of the The primary outcome was major adverse cardiac
RCRI [2]. The clinical risk factor dataset was complete events, defined as a composite of death within 30 days of
for all recruited patients and was reviewed by BB for surgery, or a serum troponin result above the upper
accuracy; part of this dataset has been used for a reference limit within the first three postoperative days.
previous publication [9]. Troponin and CRP levels were Categorical data were analysed using the Fisher’s exact
measured at some point in the 24 h before surgery. In test or Pearson’s chi-squared test where appropriate; all
April 2008, the hospital changed the troponin assay continuous data were compared using independent
from troponin T to troponin I, and in August 2008, BNP samples t-test or Mann–Whitney U-test. The statistical
390 Anaesthesia ª 2012 The Association of Anaesthetists of Great Britain and Ireland
3. Biccard et al. | Biomarkers and Holter monitoring for risk stratification Anaesthesia 2012, 67, 389–395
analysis was conducted in two stages; in the first stage, a Results
univariate analysis was conducted for all the patients who Nine hundred and seventy-eight patients were eligible
had both pre-operative biomarker and Holter monitor- for the study in the 3-year-period, of whom 788 patients
ing data. All risk factors with a univariate association of consented. The compositions of the cohorts are shown
p < 0.1 with the study outcome were entered into in Fig. 1. The study outcome occurred in 136 out of the
multivariate analysis, using binary logistic regression. If 788 recruited patients (17%, 95% CI 15–20%) and was
none of the Holter risk factors were found to be similar between the four cohorts: pre-operative troponin
independent predictors of the study outcome, then the cohort (98 ⁄ 534, 18%, 95% CI 15–22%); pre-operative
second stage of the analysis consisted of a multivariate BNP cohort (65 ⁄ 403, 16%, 95% CI 13–20%), pre-
analysis using the larger biomarker cohort. By only operative CRP cohort (87 ⁄ 508, 17%, 95% CI 14–20%);
including univariate predictors with p < 0.1 into a and pre-operative Holter cohort (58 ⁄ 318, 18%, 95% CI
subsequent multivariate regression, the events per var- 14–23).
iable ratio were kept above ten, thus minimising the bias The baseline patient characteristics are shown in
associated with the estimate of risk [12]. A backward Table 1. Patients who sustained major adverse cardiac
stepwise modelling technique was also performed, based events were older, had significantly more ischaemic
on likelihood ratios with entry and removal probabilities heart disease, diabetes and pre-operative troponin level
set at 0.05 and 0.1, respectively. For biomarker analysis, above the upper reference limit, and they also had
categorical data were used. Positive pre-operative tropo- significantly higher pre-operative BNP and CRP levels.
nin levels were defined as above the upper reference level, They were also taking significantly more beta-blocker
and positive pre-operative BNP and CRP levels were and aspirin therapy pre-operatively.
defined as above the optimal discriminatory point Univariate analysis of the Holter cohort is shown in
determined using a receiver-operating characteristic Table 2; the only Holter variable associated with
curve for the study outcome. major adverse cardiac events with a p < 0.1 was the
Finally, to determine whether any of the indepen-
dent predictors identified in the logistic regression
significantly improved pre-operative risk prediction for
postoperative major adverse cardiac events, a category- Eligible patients (n = 978)
free net reclassification was used. This reclassification
method is deemed the most objective statistical tool for
evaluating the prognostic performance of a risk predic-
Recruited patients (n = 788)
tor. The results from a category-free net reclassification Clinical risk factors (n = 788)
are independent of the clinical risk stratification tool
used during the study, and so can be used for objective
comparisons with potential future risk predictors [13].
Patients were reclassified into a high-(positive indepen- Preoperative biomarker cohort
dent risk predictors) or low-(negative independent risk Troponins (n = 560)
predictors) risk category. The success of this reclassifi-
BNP (n = 403)
cation is described by the change in net reclassification,
where a positive change reflects an improvement in risk CRP (n = 508)
stratification. Net reclassification is the difference
between the proportion of patients correctly and incor-
rectly reclassified [8] according to the study outcome.
Holter cohort (n = 318)
SPSS 15.0 for Windows (IBM, NY, USA), EpiCalc 2000
(Version 1.2, Brixton Health, UK) and SAS Software 9.1
(SAS Institute Inc., Cary, NC, USA) were used for data Figure 1 Flow diagram of recruited patients. BNP, brain
analysis. natriuretic peptide; CRP, C-reactive protein.
Anaesthesia ª 2012 The Association of Anaesthetists of Great Britain and Ireland 391
4. Anaesthesia 2012, 67, 389–395 Biccard et al. | Biomarkers and Holter monitoring for risk stratification
Table 1 Baseline patient characteristics. Values are mean (SD) number (proportion), or median [IQR (range)].
Total cohort Patients with Patients without
(n = 788) MACE (n = 136) MACE (n = 652) p value
Clinical risk factors
Age 58.2 (14.2) 62.4 (13.4) 57.4 (14.2) < 0.001
Men 512 (65%) 80 (59%) 432 (66%) 0.11
Ischaemic heart disease 275 (35%) 74 (54%) 201 (31%) < 0.001
Diabetes 338 (43%) 78 (57%) 260 (40%) < 0.001
Cardiac failure 37 (5%) 11 (8%) 26 (4%) 0.46
Cerebrovascular accident 159 (20%) 23 (17%) 136 (21%) 0.35
Creatinine > 177 lmol.l)1 18 ⁄ 730 (3%) 4 ⁄ 130 (3%) 14 ⁄ 600 (2%) 0.62
Pre-operative medications
Beta-blockers 267 (34%) 82 (60%) 185 (28%) < 0.001
Statins 685 (87%) 125 (92%) 560 (86%) 0.07
Aspirin 714 (91%) 131 (96%) 583 (89%) 0.009
Pre-operative biomarkers
Troponin I elevation 25 ⁄ 509 (5%) 20 ⁄ 98 (20%) 5 ⁄ 436 (1%) < 0.001
BNP; pg.ml)1 33.6 [12.5–93.8 76.8 [39.4–337.7 28.7 [11.1–74.7 < 0.001
(2.1–3893.0)] (4.5–3893.0)] (2.1–3138.0)]
CRP; g.dl)1 19 [5.4–67 28 [8–108 17.3 [5.0–62.2 0.008
(0–263.0)] (0.1–263.0)] (0–210.0)]
BNP, brain natriuretic peptide; CRP, C-reactive protein; MACE major adverse cardiac events.
pre-operative overnight mean heart rate. Pre-operative larger biomarker cohort was used for the subsequent
BNP and troponin elevation also had an association of logistic regression of independent predictors associated
p < 0.1 for postoperative major adverse cardiac events. with major adverse cardiac events.
However, on multivariate analysis, the mean pre-oper- The univariate associations of the biomarker cohorts
ative heart rate was removed from the model, leaving with major adverse cardiac events are shown in Table 3.
only pre-operative troponin and BNP elevation inde- The RCRI, pre-operative BNP, CRP and troponin
pendently associated with the outcome. As a result, the elevation were entered into the multivariate analysis,
with pre-operative troponin (OR 57, 95% CI 6–496,
Table 2 Univariate predictors of major adverse cardiac p < 0.001) and BNP (OR 6.0, 95% CI 2.7–12.9,
events in the Holter cohort. p < 0.001) elevation being the only independent risk
factors associated with major adverse cardiac events.
OR (95% CI) p value
The pre-operative BNP optimal cut-off had an area
Clinical risk factors
RCRI score 1.3 (1.0–1.7) 0.11 under the curve of 0.69 (95% CI 0.62–0.75, p = 0.035).
Holter risk predictors
Episodes of ST depression 1.0 (0.9–1.2) 0.66 Table 3 Univariate predictors of major adverse cardiac
Mean heart rate 1.02 (0.99–1.04) 0.09
events in the biomarker cohort.
Maximum heart rate 1.00 (0.99–1.02) 0.58
Longest duration 1.00 (1.00–1.00) 0.12
above 100 beats.min)1 OR (95% CI) p value
Pre-operative biomarkers Clinical risk factors
BNP; pg.ml)1 1.00 (1.00–1.00) < 0.001 RCRI score 1.5 (1.2–1.8) < 0.001
BNP above optimal 4.6 (2.1–10.0) < 0.001 Pre-operative biomarkers
discriminatory point BNP above optimal 5.0 (2.7–9.4) < 0.001
of 48.1 pg.ml)1 discriminatory point
CRP; g.dl)1 1.01 (0.99–1.01) 0.12 of 39.4 pg.ml)1
CRP above optimal 1.7 (0.9–3.4) 0.13 CRP above the optimal 1.8 (1.1–2.9) 0.012
discriminatory point discriminatory point
of 22 g.dl)1 of 22 g.dl)1
Troponin I > 0.1 ng.ml)1 38.3 (4.6–320.0) 0.001 Troponin I > 0.1 ng.ml)1 22.1 (8.1–60.0) < 0.001
BNP, brain natriuretic peptide; bpm, beats.min)1; CRP, C- BNP, brain natriuretic peptide; CRP, C-reactive protein; RCRI,
reactive protein; RCRI, revised cardiac risk index. revised cardiac risk index.
392 Anaesthesia ª 2012 The Association of Anaesthetists of Great Britain and Ireland
5. Biccard et al. | Biomarkers and Holter monitoring for risk stratification Anaesthesia 2012, 67, 389–395
Table 4 Net reclassification change for postoperative major adverse cardiac events using pre-operative BNP and tro-
ponins.
Reclassification change
of patients without Reclassification change
MACE of patients with MACE Overall net reclassification change
Cohort proportion p-value proportion p-value proportion p value
BNP above the optimal +21% < 0.0001 +54% < 0.0001 +75% (95% CI 52–98%) < 0.0001
discriminatory point
Troponin above the +98% < 0.0001 )59% < 0.0001 +39% (95% CI 22–55%) 0.0006
upper reference limit
BNP, brain natriuretic peptide; MACE, major adverse cardiac event.
The reclassification improvement for postoperative major adverse cardiac event by classifying many of them
major adverse cardiac events associated with pre-oper- as low-risk. As this would have significant negative
ative troponin and BNP elevation is shown in Table 4. clinical impact, we do not advocate using pre-operative
Both pre-operative troponin and BNP analysis signifi- troponins as a screening test to exclude high-risk
cantly improved overall risk stratification. However, patients. In contrast, pre-operative BNP was an inde-
pre-operative troponin risk stratification significantly pendent predictor of major adverse cardiac events and
decreased correct classification of patients who devel- significantly improved the risk stratification of patients
oped major adverse cardiac events. with and without major adverse cardiac events.
There are a number of potential reasons why pre-
Discussion operative BNP may be a better predictor of postoper-
This study has shown that, in elective vascular surgical ative cardiac complications than troponins. Brain
patients, only pre-operative BNP and troponins have the natriuretic peptide is rapidly secreted from ventricular
capacity to change risk prediction significantly. Impor- myocytes when exposed to even minor elevations in
tantly, this ability to risk-stratify is independent of the ventricular pressure or volume loading [18], or from
pre-operative clinical risk factors found in the RCRI, a myocardial ischaemia [18, 19]. This allows pre-operative
finding consistent with earlier work on pre-operative BNP elevation to identify a vulnerable ventricle at risk of
BNP in vascular patients [14]. As a result of the RCRI’s a major adverse event. In contrast, troponin elevations,
poor ability to risk-stratify vascular surgical patients [15, as detected by standard sensitivity troponin assays, most
16], it is important to identify whether alternative risk commonly reflect myocyte necrosis as the final common
stratification tools (i.e. pre-operative ECG and pre- pathway of a damaged ventricle. Pre-operative troponin
operative biomarkers) are appropriate for this popula- elevation probably reflects a ventricle that is too far
tion. down the pathway of cardiovascular injury to provide
We statistically defined a risk factor with clinical further clinically useful pre-operative risk stratification
utility as one that was both independently associated information. This is further emphasised by the pattern
with the outcome and also significantly improved pre- of troponin increase commonly seen in the postopera-
operative risk category classification for subsequent tive period, where only 12% of patients who sustain a
major adverse cardiac events [7, 17]. Adopting the risk peri-operative myocardial infarction have troponin
predictors identified in this study should result in a elevation on postoperative day 1, while 77% have
significant change in pre-operative risk categorisation, troponin elevation by day 3 [20]. Similarly, in our
which could potentially alter pre-operative clinical study, only 4.7% of patients had pre-operative troponin
management. elevation, compared with 15.9% with postoperative
Pre-operative troponins significantly improved pa- troponin elevation. These findings demonstrate that
tient risk reclassification overall, but worsened the pre-operative troponin elevation is poorly associated
reclassification of the sub-cohort of patients who had a with postoperative cardiac events.
Anaesthesia ª 2012 The Association of Anaesthetists of Great Britain and Ireland 393
6. Anaesthesia 2012, 67, 389–395 Biccard et al. | Biomarkers and Holter monitoring for risk stratification
With regards to the Holter monitoring used for this Acknowledgements
study, only two channels (modified V2 and V5) were We thank Tecmed South Africa for providing us with
analysed. This is certainly not equivalent to the 12-lead Holter hardware and software at cost price. The study
ECG monitoring used in other studies of peri-operative itself was funded through a grant from the Medical
myocardial ischaemia, and may have decreased the Research Council of South Africa, awarded to BB.
sensitivity of our Holter data. However, we do not
consider this to be a significant limitation, as the Competing interests
combination of two precordial leads has a reported No competing interests declared.
sensitivity of over 90% for the detection of myocardial
ischaemia [21]. References
This is the first study to compare the performance of 1. Devereaux PJ, Goldman L, Cook DJ, Gilbert K, Leslie K, Guyatt GH.
Perioperative cardiac events in patients undergoing noncardiac
three different pre-operative risk stratification tech- surgery: a review of the magnitude of the problem, the
niques (i.e., clinical risk predictors, biochemical risk pathophysiology of the events and methods to estimate and
communicate risk. Canadian Medical Association Journal 2005;
markers and ECG Holter monitoring) simultaneously. 173: 627–34.
These findings have important implications for the 2. Lee TH, Marcantonio ER, Mangione CM, et al. Derivation and
development of an evidence-based clinical approach to prospective validation of a simple index for prediction of cardiac
risk of major noncardiac surgery. Circulation 1999; 100: 1043–9.
risk prediction in vascular surgical patients, and warrant 3. Mangano DT, Browner WS, Hollenberg M, London MJ, Tubau JF,
verification in an independent cohort. Tateo IM. Association of perioperative myocardial ischemia with
cardiac morbidity and mortality in men undergoing noncardiac
We can now summarise our understanding of the surgery. The Study of Perioperative Ischemia Research Group.
role of BNP in pre-operative risk stratification for New England Journal of Medicine 1990; 323: 1781–8.
vascular surgical patients as follows: (1) BNP improves 4. Choi JH, Cho DK, Song YB, et al. Preoperative NT-proBNP and CRP
predict perioperative major cardiovascular events in noncardiac
RCRI risk stratification [9]; (2) BNP is superior to and surgery. Heart 2010; 96: 56–62.
independent of the RCRI and its individual compo- 5. Fleisher LA, Beckman JA, Brown KA, et al. ACC ⁄ AHA 2007
Guidelines on Perioperative Cardiovascular Evaluation and Care
nents [10]; and (3) BNP remains independently for Noncardiac Surgery: Executive Summary: A Report of the
predictive even in the presence of pre-operative American College of Cardiology ⁄ American Heart Association Task
troponin evaluation and myocardial ischaemia Force on Practice Guidelines (Writing Committee to Revise the
2002 Guidelines on Perioperative Cardiovascular Evaluation for
monitoring. Noncardiac Surgery): Developed in Collaboration With the
Based on the recommended process for the critical American Society of Echocardiography, American Society of
Nuclear Cardiology, Heart Rhythm Society, Society of Cardiovas-
evaluation of novel biomarkers of cardiovascular risk cular Anesthesiologists, Society for Cardiovascular Angiography
before integration in clinical practice, a progressive six- and Interventions, Society for Vascular Medicine and Biology, and
phase evaluation has been recommended [22]. Brain Society for Vascular Surgery. Circulation 2007; 116: 1971–96.
6. Poldermans D, Bax JJ, Boersma E, et al. Guidelines for pre-
natriuretic peptide evaluation has now reached the operative cardiac risk assessment and perioperative cardiac
fourth stage (that of clinical utility). As it has been management in non-cardiac surgery: the Task Force for Preop-
erative Cardiac Risk Assessment and Perioperative Cardiac
shown significantly to outperform the RCRI as well as Management in Non-cardiac Surgery of the European Society
other methods, we believe that we have now established of Cardiology (ESC) and European Society of Anaesthesiology
BNP as the modality of choice for estimating risk in (ESA). European Heart Journal 2009; 30: 2769–812.
7. Cook NR. Statistical evaluation of prognostic versus diagnostic
these patients. models: beyond the ROC curve. Clinical Chemistry 2008; 54:
We propose that future research into risk stratifi- 17–23.
8. Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS.
cation for vascular surgical patients should include the Evaluating the added predictive ability of a new marker: from
verification of these results in an independent vascular area under the ROC curve to reclassification and beyond.
surgical cohort and determination of whether pre- Statistics in Medicine 2008; 27: 157–72; discussion 207–12.
9. Biccard BM, Naidoo P. The role of brain natriuretic peptide in
operative management of vascular surgical patients prognostication and reclassification of risk in patients undergo-
based on pre-operative BNP levels improves clinical ing vascular surgery. Anaesthesia 2011; 66: 379–85.
10. Biccard BM, Lurati Buse GA, Burkhart C, et al. The influence of
outcome (stage 5 of the evaluation of novel biomarkers) clinical risk factors on pre-operative B-type natriuretic peptide
[22]. These findings should also be evaluated in non- risk stratification of vascular surgical patients. Anaesthesia
vascular, non-cardiac surgery. 2012; 67: 55–59.
394 Anaesthesia ª 2012 The Association of Anaesthetists of Great Britain and Ireland
7. Biccard et al. | Biomarkers and Holter monitoring for risk stratification Anaesthesia 2012, 67, 389–395
11. Rodseth RN, Padayachee L, Biccard BM. A meta-analysis of the 16. Biccard BM, Rodseth RN. Utility of clinical risk predictors for
utility of pre-operative brain natriuretic peptide in predicting preoperative cardiovascular risk prediction. British Journal of
early and intermediate-term mortality and major adverse Anaesthesia 2011; 107: 133–43.
cardiac events in vascular surgical patients. Anaesthesia 2008; 17. Ray P, Le Manach Y, Riou B, Houle TT. Statistical evaluation of a
63: 1226–33. biomarker. Anesthesiology 2010; 112: 1023–40.
12. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A 18. Rodseth RN. B type natriuretic peptide-a diagnostic break-
simulation study of the number of events per variable in logistic through in peri-operative cardiac risk assessment? Anaesthesia
regression analysis. Journal of Clinical Epidemiology 1996; 49: 2009; 64: 165–78.
1373–9. 19. Struthers A, Lang C. The potential to improve primary prevention
13. Pencina MJ, D’Agostino RB Sr, Steyerberg EW. Extensions of net in the future by using BNP ⁄ N-BNP as an indicator of silent
reclassification improvement calculations to measure usefulness ‘pancardiac’ target organ damage: BNP ⁄ N-BNP could become
of new biomarkers. Statistics in Medicine 2011; 30: 11–21. for the heart what microalbuminuria is for the kidney. European
14. Rodseth RN, Lurati Buse GA, Bolliger D, et al. The predictive Heart Journal 2007; 28: 1678–82.
ability of pre-operative B-type natriuretic peptide in vascular 20. Biccard BM, Rodseth RN. The pathophysiology of peri-operative
patients for major adverse cardiac events an individual patient myocardial infarction. Anaesthesia 2010; 65: 733–41.
data meta-analysis. Journal of the American College of Cardi- 21. Landesberg G. Monitoring for myocardial ischemia. Best Practice
ology 2011; 58: 522–9. and Research Clinical Anaesthesiology 2005; 19: 77–95.
15. Ford MK, Beattie WS, Wijeysundera DN. Systematic review: 22. Hlatky MA, Greenland P, Arnett DK, et al. Criteria for evaluation
prediction of perioperative cardiac complications and mortality of novel markers of cardiovascular risk: a scientific statement
by the revised cardiac risk index. Annals of Internal Medicine from the American Heart Association. Circulation 2009; 119:
2010; 152: 26–35. 2408–16.
Anaesthesia ª 2012 The Association of Anaesthetists of Great Britain and Ireland 395