Morbidity following CABG (coronary artery bypass grafting) is difficult to predict and leads to increased healthcare costs. We hypothesized that pre-operative CMR (cardiac magnetic resonance) findings would predict resource utilization in elective CABG. Over a 12-month period, patients requiring elective CABG were invited to undergo CMR 1 day prior to CABG. Gadolinium-enhanced CMR was performed using a trueFISP inversion recovery sequence on a 1.5 tesla scanner (Sonata; Siemens). Clinical data were collected prospectively. Admission costs were quantified based on standardized actual cost/day. Admission cost greater than the median was defined as ‘increased’. Of 458 elective CABG cases, 45 (10%) underwent pre-operative CMR. Pre-operative characteristics [mean (S.D.) age, 64 (9) years, mortality (1%) and median (interquartile range) admission duration, 7 (6–8) days] were similar in patients who did or did not undergo CMR. In the patients undergoing CMR, eight (18%) and 11 (24%) patients had reduced LV (left ventricular) systolic function by CMR [LVEF (LV ejection fraction) <55%] and echocardiography respectively. LE (late enhancement) with gadolinium was detected in 17 (38%) patients. The average cost/day was $2723. The median (interquartile range) admission cost was$19059 ($10891–157917). CMR LVEF {OR (odds ratio), 0.93 [95% CI (confidence interval), 0.87–0.99]; P=0.03} and SV (stroke volume) index [OR 1.07 (95% CI, 1.00–1.14); P=0.02] predicted increased admission cost. CMR LVEF (P=0.08) and EuroScore tended to predict actual admission cost (P=0.09), but SV by CMR (P=0.16) and LV function by echocardiography (P=0.95) did not. In conclusion, in this exploratory investigation, pre-operative CMR findings predicted admission duration and increased admission cost in elective CABG surgery. The cost-effectiveness of CMR in risk stratification in elective CABG surgery merits prospective assessment. ## INTRODUCTION Quality of care and resource utilization have increasing importance in cardiovascular healthcare provision [1]. CABG (coronary artery bypass grafting) is an effective but expensive treatment [25]. Despite low mortality rates following elective CABG [2,6], post-operative morbidity is fairly common, leading to prolonged hospital admission and increased resource utilization [79]. Hospitalization costs are the leading cause of direct healthcare expenditure [3,5,10]. In cardiac surgery, predicting prolonged admission is notoriously difficult [11,12], and risk assessment tools for morbidity [13] and mortality [12] place an emphasis on unstable high-risk characteristics. Consequently, prediction of admission duration and costs may be less reliable in lower-risk patients, who represent the majority of CABG cases [2]. CMR [cardiac MR (magnetic resonance)] has high tissue contrast and spatial resolution, and has become the gold standard technique for assessment of cardiac structure and function [1416]. Despite widespread availability, pre-operative CMR is usually not requested, and this is probably because of conventional reliance on echocardiography. However, compared with CMR, echocardiography is less accurate and provides no information on viability, and acoustic limitations are fairly common [14]. Information about the use of CMR for the prediction of in-hospital outcomes in elective (low-risk) CABG is lacking. We hypothesized that CMR findings prior to CABG would correlate with post-operative morbidity and predict resource utilization. Specifically, in the present study, we investigated the following questions. First, could LE (late enhancement) with gadolinium be feasible as part of a routine assessment in low-risk patients prior to elective CABG? Secondly, which characteristics, including those obtained from CMR, correlate with in-hospital outcomes? Thirdly, do CMR findings predict resource utilization? To answer these questions, CMR was performed in a representative subset of patients undergoing elective CABG over a 12-month period. ## MATERIALS AND METHODS Patients with multivessel CAD (coronary artery disease) undergoing elective CABG were invited to participate and gave informed consent. Patients undergoing CABG combined with valve surgery were not included. There were no other exclusion criteria. One to two CMR studies took place per week over a 12-month period. Limitations in MR scanner availability due to other research activity precluded inclusion of more patients. MR data were not intended to guide clinical management. Clinical data were obtained from hospital electronic databases. Hospital costs were quantified based on actual standardized Scottish National Health Service (2005) costs per cardiac surgical bed day. This study was approved by the North Glasgow University Hospitals NHS Trust Ethics Committee. ### CMR image acquisition: pre-gadolinium contrast imaging CMR was performed with a 1.5 tesla scanner (Sonata; Siemens) 1 day prior to CABG according to methods reported previously [17]. Fast imaging at steady-state free precession cines (TrueFISP by Siemens) were used throughout for all pre-gadolinium cines. The images were first acquired in a long-axis plane [vertical long-axis, horizontal long-axis and LV (left ventricular) outflow tract], followed by sequential short-axis LV cine loops from the atrioventricular ring to the LV apex. Pre-gadolinium imaging parameters, which were standardized for all subjects, included: repetition time, 3.2 ms; echo time, 1.6 ms; flip angle, 20°; field of view, 276 mm×340 mm; and pixel dimensions, 2.0 mm×1.3 mm. Imaging slices (8 mm) with a 2 mm interslice gap were used. ### CMR image analysis: pre-gadolinium LV images were analysed using dedicated software (Argus VA60C 2004; Siemens). Individual scans were coded by number and analysed in batches. Investigators were blinded to patient characteristics and in-hospital outcomes following CABG. LV volumes [LVEDV (LV end-diastolic volume) and LVESV (LV end-systolic volume)] were determined by manual planimetry of selected short-axis images, as described previously [1719]. Particular methodological points of note included the deliberate inclusion of trabeculations and papillary muscles in all analyses [19]. The most basal ventricular slice was identified visually and then cross-referenced with the horizontal long-axis view before its inclusion in the final analysis to ensure this slice was indeed ventricular and not atrial. Two observers evaluated LV dimensions and function. In a subset of ten patients, the inter-observer correlations were: r=0.87 (P=0.002) for LVEDV, r=0.86 (P=0.003) for LVESV, r=0.83 (P=0.005) for SV (stroke volume) and r=0.85 (P=0.004) for LV mass. According to CMR, a reduced LVEF (LV ejection fraction) was defined as <55% [20]. LV dimensions were indexed to body surface area [21]. ### Gadolinium-contrast-enhanced CMR image acquisition All patients underwent dynamic gadolinium-contrast-enhanced CMR. Gadolinium imaging was performed by injection of gadolinium diethylene triaminepenta-acetic acid at a dose of 0.2 mmol/kg of body weight, followed 10 min later by short-axis stack image acquisition with a gadolinium-sensitive inversion recovery technique [inversion recovery TurboFLASH (fast low-angle shot); Siemens] and identical slice positions as in the pre-gadolinium cines [22]. Standardized scan settings were used for CMR in all subjects as follows: repetition time, 11.6 ms; echo time, 4.3 ms; flip angle, 20°; pixel dimensions, 2.2 mm×1.3 mm×8 mm; number of segments, 23. Imaging slices (8 mm) with a 2 mm interslice gap were used. The TurboFLASH inversion time was optimized on an individual patient basis. Successful nulling of normal myocardium was deemed to have been achieved once the LV myocardium appeared black and homogenous. An inversion time between 240 and 280 ms was required to achieve this. Artefacts on the LE images were excluded by the acquisition of ‘swapped phase’ images through slice planes, which appeared to demonstrate intramyocardial gadolinium enhancement. This technique involves the ‘swapping’ of the phase-encoding and frequency-encoding directions, as the phase-encoding direction is particularly prone to artefacts caused by cardiac or chest wall movement. ### Gadolinium-contrast-enhanced CMR image analyses Our approach to methods of LE analyses have been described in detail recently [23]. Briefly, LE was defined as an area of visually identified gadolinium enhancement with a mean SI (signal intensity) that was more than 1 S.D. higher than the mean SI of an adjacent area of reference myocardium, which, although nulled, had a mean SI significantly above zero. LE volume was determined by planimetry of any areas of gadolinium enhancement meeting these criteria. LE mass was calculated by multiplying gadolinium enhancement volume by myocardial density (1.05 g/cm3), and these results were used in the subsequent correlations and discussion as the absolute measured value of gadolinium enhancing tissue. Regional LV hypoperfusion on first-pass was visually assessed and documented using a 16-segment model. An FPS (first-pass perfusion score) was determined according to the method of Baks et al. [24]. Each segment was evaluated and scored as (1) normal enhancement, (2) <30% wall thickness subendocardial hypoenhancement or (3) >30% hypoenhancement. Infarct size was defined as an area of LE. PMO (persistent microvascular obstruction) was defined as an area of reduced signal intensity within an area of LE [2527]. Infarct size and PMO were quantified on the three-dimensional volume by manual delineation of the enhanced and unenhanced myocardium with different contours (Argus software; Siemens) and is expressed as a percentage of LV mass [23,25,26]. PMO was included in the infarct size quantification. ### Echocardiography Echocardiograms were performed by experienced technicians and reports were issued by cardiologists. Reduced LV systolic function was defined as an LVEF <0.40 or a qualitative report of depressed LV systolic function (mild, moderate or severe), which is the standard reporting method in our service [28]. ECGs were performed within 2 months of surgery. ### Statistical analysis We hypothesized that subtle abnormalities in LV function detected by CMR would be associated with prolonged admission duration >7 days. With 22 subjects in each group (admission duration, normal compared with prolonged), a mean between-group difference in LVEF of 10% (S.D. of 10%) could be detected with 90% power assuming a Type I error of 0.05. The distribution of all continuous data were visually inspected on a normal plot. Normality was confirmed or excluded using the Shapiro–Francia test. Mean (S.D.) values and medians (interquartile range) were calculated for normally and non-normally distributed data respectively. Correlations between normally and non-normally distributed variables were tested by Pearson's or Spearman's rho tests respectively. All tests were two-tailed. Increased admission duration was taken to represent 1 week or more, consistent with the median duration of admission, which was 7 days. For resource utilization, the total admission cost was calculated by multiplying the average cost/day for cardiac surgery in the Western Infirmary (2005) by individualized length of stay. A currency exchange rate of £1 to$1.90 was used to estimate costs in US$. Predictors of admission cost and increased admission cost were assessed using simple and logistic regression models, including in the subset of patients with LE (n=17). For the univariate models, the CV (coefficient of variation) (for continuous data) and OR (odds ratio) (for categorical data) for a given increment in the covariate are reported along with 95% CIs (confidence intervals) and the associated P value. A significance level of 5% was used in all tests. No adjustment was made to P values to account for multiple testing. All statistical analyses were performed using STATA version 7 (Statacorp). ## RESULTS ### All patients A total of 458 elective CABG cases were performed in our centre between 7 April 2004 and 19 April 2005. The characteristics of these patients are shown in Table 1. Previous MI (myocardial infarction) and PCI (percutaneous coronary intervention) were more frequent in the CMR group, whereas operative risk tended to be higher in the non-CMR group. The number of grafts, admission duration and mortality were similar (Table 1). Table 1 Clinical characteristics of patients undergoing elective CABG according to pre-operative CMR status Values are means (S.D.), medians (interquartile range), or numbers. BMI, body mass index; NYHA, New York Heart Association. Clinical characteristicPre-operative CMR (n=45)No pre-operative CMR (n=413)P value Age (years) 63 (9) 65 (9) 0.16 BMI (kg/m229 (4) 29 (4) 0.50 Male gender (n37 (82%) 319 (80%) 0.7 History of MI (n24 (53%) 164 (40%) 0.08 Diabetes (n Non-insulin treated 3 (7%) 21 (5%) 0.64 Insulin-treated 9 (20%) 64 (15%) Hypertension (n31 (69%) 266 (64%) 0.55 Chronic lung disease (n2 (4%) 47 (11%) 0.15 Peripheral vascular disease (n1 (2%) 39 (9%) 0.10 History of PCI (n5 (11%) 11 (3%) 0.003 Serum creatinine (μmol/l) 102 (95–113) 102 (93–116) 0.88 Angina grade (Canadian Cardiovascular Class 0/1/2/3/4) (n1 (2%)/9 (20%)/20 (44%)/13 (29%)/2 (4%) 20 (5%)/52 (13%)/176 (43%)/128 (31%)/37 (9%) 0.50 NYHA functional class I/II/III (n14 (31%)/22 (49%)/9 (20%) 95 (23%)/192 (46%)/113 (27%)/13 (3%) 0.33 Coronary artery grafts (n) (1–2/3/4–5) 15 (33%)/25(56%)/5 (11%) 88 (21%)/268 (65%)/57 (14%) 0.2 EuroScore (%) 2.5 (1.6) 3.2 (2.3) 0.03 Parsonnet risk (%) 5 (6) 6 (7) 0.11 Drug therapy (n Aspirin 37 (82%) 362 (88) 0.30 β-Blocker 39 (87%) 334 (81) 0.34 ACE inhibitor 25 (44%) 223 (54) 0.84 Clinical characteristicPre-operative CMR (n=45)No pre-operative CMR (n=413)P value Age (years) 63 (9) 65 (9) 0.16 BMI (kg/m229 (4) 29 (4) 0.50 Male gender (n37 (82%) 319 (80%) 0.7 History of MI (n24 (53%) 164 (40%) 0.08 Diabetes (n Non-insulin treated 3 (7%) 21 (5%) 0.64 Insulin-treated 9 (20%) 64 (15%) Hypertension (n31 (69%) 266 (64%) 0.55 Chronic lung disease (n2 (4%) 47 (11%) 0.15 Peripheral vascular disease (n1 (2%) 39 (9%) 0.10 History of PCI (n5 (11%) 11 (3%) 0.003 Serum creatinine (μmol/l) 102 (95–113) 102 (93–116) 0.88 Angina grade (Canadian Cardiovascular Class 0/1/2/3/4) (n1 (2%)/9 (20%)/20 (44%)/13 (29%)/2 (4%) 20 (5%)/52 (13%)/176 (43%)/128 (31%)/37 (9%) 0.50 NYHA functional class I/II/III (n14 (31%)/22 (49%)/9 (20%) 95 (23%)/192 (46%)/113 (27%)/13 (3%) 0.33 Coronary artery grafts (n) (1–2/3/4–5) 15 (33%)/25(56%)/5 (11%) 88 (21%)/268 (65%)/57 (14%) 0.2 EuroScore (%) 2.5 (1.6) 3.2 (2.3) 0.03 Parsonnet risk (%) 5 (6) 6 (7) 0.11 Drug therapy (n Aspirin 37 (82%) 362 (88) 0.30 β-Blocker 39 (87%) 334 (81) 0.34 ACE inhibitor 25 (44%) 223 (54) 0.84 ### CMR patients Pre-operative gadolinium-enhanced CMR was performed in 45 elective CABG patients (Table 2) with, on average, one to two MR studies per week according to scanner availability. Eight (18%) patients had reduced LVEF defined by CMR, whereas 11 (24%) patients had reduced LV systolic function according to echocardiography. Only four (9%) patients had reduced LV systolic function by both methods. LVEF was lower in men [61 (10)%] than in women [73 (6)%; P=0.002]. LE was detected in 17 (38%) patients. Six of these patients had evidence of PMO [1.3 (0.6)%], all of whom had a clinical history of MI. Table 2 Findings from dynamic gadolinium-contrast-enhanced CMR *LE was present in 17 patients, in whom six had evidence of PMO. More patients with a clinical history of MI had evidence of LE [11 (8)%] compared with patients without a history of MI [3.9 (3.8)%; P=0.04]. Mean (S.D.) [interquartile range] LVEF (%) 64 (10) [34–87] LVEDV index (ml/m272 (18) [39–114] LVESV index (ml/m227 (12) [7–62] SV index (ml/m245 (11) [26–66] LV mass (g/m278 (22) [39–125] First-pass perfusion defect (n9 (20%) First-pass perfusion score (n 16 36 (80%) 18–21 4 (8%) 22–25 5 (12%) Patients with LE (n17 (38%) Territories with transmural extent of infarction >50% (n 0 1 11 (24%) 2 3 (7%) Infarct size (% LV mass)* 10 (8) [1.2–26.6] PMO (% LV mass) 1.3 (0.4) [0.6–1.7] Mean (S.D.) [interquartile range] LVEF (%) 64 (10) [34–87] LVEDV index (ml/m272 (18) [39–114] LVESV index (ml/m227 (12) [7–62] SV index (ml/m245 (11) [26–66] LV mass (g/m278 (22) [39–125] First-pass perfusion defect (n9 (20%) First-pass perfusion score (n 16 36 (80%) 18–21 4 (8%) 22–25 5 (12%) Patients with LE (n17 (38%) Territories with transmural extent of infarction >50% (n 0 1 11 (24%) 2 3 (7%) Infarct size (% LV mass)* 10 (8) [1.2–26.6] PMO (% LV mass) 1.3 (0.4) [0.6–1.7] #### In-hospital outcomes A 73 year-old diabetic man died on the third post-operative day. He had a reduced pre-operative LVEF (46%) and evidence of LE (infarct size, 26%; PMO, 2%). The median duration of admission was 7 (6–8) days. The minimum and maximum admission durations were 4 to 58 days. Compared with patients undergoing CMR with a normal admission duration, patients with an increased admission duration (>7 days) had a lower LVEF, a higher LVESV index and more post-operative complications (Table 3). Admission duration correlated with LVEF (R=−0.34, P=0.02) and tended to correlate with the EuroScore (R=0.22, P=0.08) and LVESV index (R=0.28, P=0.06). No other clinical or CMR findings correlated with admission duration. Table 3 Pre-operative characteristics and post-operative complications in CMR patients according to admission duration ≤ or >7 days Values are means (S.D.), medians (interquartile range), or numbers (percentage). *One patient had a stroke and one other patient died; †infection was recorded if there was septicaemia, or involvement of the leg, sternum or mediastinum. Admission duration ≤7 days>7 daysP value Pre-operative Clinical n 29 (64%) 16 (36%) Age (years) 61 (8) 65 (10) 0.17 BMI (kg/m230 (4) 28 (4) 0.21 Creatinine (μmol/l) 103 (15) 117 (38) 0.08 Haemoglobin (g/dl) 14.4 (0.9) 13.8 (0.9) 0.05 Diseased arteries (n Two 12 (41%) 5 (31%) Three 17 (58%) 11 (69%) 0.50 Estimated in-hospital mortality EuroScore (%) 2 (1–4) 3 (2–4) 0.20 Parsonnet risk (%) 4 (3–7) 5.5 (2.5–12) Cardiac CMR LVEF (%) 66 (10) 59 (10) 0.02 LVEDV index (ml/m268 (16) 78 (19) 0.09 LVESV (ml) 23 (11) 33 (12) 0.01 SV index (ml/m245 (10) 45 (12) 0.87 LV mass index (g/m280 (21) 75 (25) 0.44 First-pass perfusion score (n17 (2) 18 (3) 0.34 Territories with transmural extent of infarction >50% (n 0 20 (69%) 11 (69%) 1 8 (28%) 3 (18%) 0.45 2 1 (3%) 2 (12%) Infarct size (% LV mass) 3.5 (7.0) 4.4 (7.2) 0.70 PMO (% LV mass) 0.13 (0.44) 0.25 (0.54) 0.44 Post-operative complications Major adverse cardiac and cerebrovascular events (n)* 1 (3%) 1 (6%) 0.66 Atrial flutter/fibrillation (n4 (14%) 10 (62%) 0.001 Infection (n)† 1 (3%) 1 (6%) 0.66 Blood product transfusion (n8 (28%) 11 (69%) 0.007 Admission duration ≤7 days>7 daysP value Pre-operative Clinical n 29 (64%) 16 (36%) Age (years) 61 (8) 65 (10) 0.17 BMI (kg/m230 (4) 28 (4) 0.21 Creatinine (μmol/l) 103 (15) 117 (38) 0.08 Haemoglobin (g/dl) 14.4 (0.9) 13.8 (0.9) 0.05 Diseased arteries (n Two 12 (41%) 5 (31%) Three 17 (58%) 11 (69%) 0.50 Estimated in-hospital mortality EuroScore (%) 2 (1–4) 3 (2–4) 0.20 Parsonnet risk (%) 4 (3–7) 5.5 (2.5–12) Cardiac CMR LVEF (%) 66 (10) 59 (10) 0.02 LVEDV index (ml/m268 (16) 78 (19) 0.09 LVESV (ml) 23 (11) 33 (12) 0.01 SV index (ml/m245 (10) 45 (12) 0.87 LV mass index (g/m280 (21) 75 (25) 0.44 First-pass perfusion score (n17 (2) 18 (3) 0.34 Territories with transmural extent of infarction >50% (n 0 20 (69%) 11 (69%) 1 8 (28%) 3 (18%) 0.45 2 1 (3%) 2 (12%) Infarct size (% LV mass) 3.5 (7.0) 4.4 (7.2) 0.70 PMO (% LV mass) 0.13 (0.44) 0.25 (0.54) 0.44 Post-operative complications Major adverse cardiac and cerebrovascular events (n)* 1 (3%) 1 (6%) 0.66 Atrial flutter/fibrillation (n4 (14%) 10 (62%) 0.001 Infection (n)† 1 (3%) 1 (6%) 0.66 Blood product transfusion (n8 (28%) 11 (69%) 0.007 CMR LVEF univariately predicted increased admission duration [OR, 0.93 (95% CI, 0.87–0.99); P=0.03], and tended to predict actual admission duration [CV, −0.21 (95% CI, −0.44 to 0.03); P=0.08]. When the EuroScore data were added to this model, the prognostic value of CMR LVEF tended to be retained [OR, 0.94 (95% CI, 0.98–1.005); P=0.069], whereas the EuroScore did not predict this event (P=0.12). ### Resource utilization The average cost/day was$2723. The median admission cost was $19059 ($16336–21782). The admission cost ranged widely from a minimum of $10891 to a maximum of$157917.

LVEF by CMR predicted increased admission cost [cost>median value: OR, 0.93 (95% CI, 0.87–0.99); P=0.03] and tended to predict actual admission cost [CV, −565.65 (95% CI, −1202.48 to 71.19); P=0.08]. The EuroScore tended to predict actual admission cost (P=0.09), but SV by CMR (P=0.16) and LV function by echocardiography (P=0.95) did not.

## DISCUSSION

In this exploratory investigation, we found that CMR findings correlated with admission duration and predicted resource utilization in elective CABG. To the best of our knowledge, this study is the first of its kind.

### Clinical relevance

Our findings are interesting for several reasons. First, CMR took place 1 day prior to elective CABG. CMR was feasible (scan time <45 min) and uncomplicated. Secondly, despite the small sample size and low rate of post-operative complications, CMR findings, and in particular LV function, correlated with admission duration. Notably, other than the EuroScore, no other variable correlated with admission duration. CMR LVEF retained its prognostic value, even when combined in a regression model with the EuroScore. Thirdly, based on actual in-hospital costs, CMR predicted increased resource utilization. Our observations raise the possibility that CMR could be useful for risk assessment in lower-risk stable patients referred for CABG, with potential economic benefits [29].

Patients with a prolonged admission following CABG had depressed pre-operative cardiac function, as revealed by CMR. Furthermore, these patients experienced an increased frequency of post-operative cardiovascular complications, including atrial fibrillation, which probably contributed to their prolonged admission duration. Although CMR was performed in a subset of patients, the characteristics and in-hospital outcomes of these patients were broadly similar to those of all patients who underwent elective CABG during the study period.

For CMR to be a useful risk prediction instrument, it should be performed earlier than was done in the present study. CMR might take place once the decision for CABG has been taken. Potentially, CMR could be performed with a view to triage of risk, which in turn could help plan for bed availability and therefore guide the operation date. Alternatively, CMR could be performed much earlier in the patient journey. As stress perfusion CMR can inform on the presence of regional ischaemia, as well as LV function, CMR could be used as the primary method of cardiac imaging to help guide the decision regarding revascularization, whilst also informing on surgical risk. Future prospective studies will be required to evaluate these different approaches.

### Current limitations for prediction of morbidity in low-risk CABG

Unlike prediction of mortality [30], prediction of non-fatal post-operative complications is notoriously difficult [11,12]. Well-validated risk prediction instruments for in-hospital mortality, such as the EuroScore [30], may not predict post-operative morbidity reliably [12]. This is particularly the case in lower-risk stable patients who represent the greatest proportion of CABG cases. In the present study, the EuroScore correlated weakly with admission duration.

Prolonged ICU (intensive care unit) bed occupancy leads to substantially increased costs, including indirect financial losses through blocked beds and reduced surgical activity. Uncertainty exists about which patients may experience prolonged admissions and, despite the availability of risk tools to predict prolonged length of stay [13,3134], they are not consistently used to guide management. One reason for the limited utility of current risk instruments is the strong weighting accorded to emergency surgery [13]. Importantly, the absolute number of post-operative complications is probably greatest in lower-risk patients, in whom risk instruments will have the least positive predictive value.

Our present study suggests CMR may represent a sensitive method for pre-operative risk prediction. Indeed, the fact that no other traditional prognostic characteristics, including LV function by echocardiography [12,13,33], predicted admission duration, raises the possibility that CMR may be a more sensitive risk predictor in elective CABG than existing methods. Echocardiography has a lower accuracy than CMR for the assessment of LV systolic function [14,29], which may explain the lack of correlation of LV function by echocardiography with admission duration.

### CABG, resource utilization and the potential use of CMR for risk prediction

Although morbidity and mortality rates in our patients were low, admission duration was >1 week in over half of the patients. Consequently, admission costs ranged widely. In the U.K., the average cost/CABG in 2003 was $17596 (Scotland), and the national in-hospital costs for CABG were approx.$500 million [5]. In 2003, expenditure in the U.S.A. for in-hospital care was $94.1 billion, of which$60 (64%) billion was for CAD [35]. In 2005, CABG was performed in 145333 isolated cases, and the mean post-procedure length of stay was 6.9 days [3]. Average CABG costs in selected hospitals in the U.S.A. and Canada (1997–2001) were $20673 and$10373 respectively (P<0.001) [4].

In this context, the incremental unit cost of CMR compared with echocardiography (approx. \$1000 [14]) is modest. Consequently, pre-operative imaging and risk assessment by CMR has the potential to be a cost-effective strategy should CMR-based management lead to reductions in peri-operative complications. Importantly, CMR also provides quantitative information on reversible myocardial ischaemia, viability, infarct size and thoracic pathology, all of which may be useful in the evaluation and management of patients being considered for revascularization.

### Conclusions

LV function measured by CMR prior to elective CABG correlates with admission duration and predicts increased resource utilization. CMR may be a more sensitive tool for risk assessment than conventional approaches in this setting. Since pre-operative cardiac imaging is mandatory, we suggest the modest additional cost of CMR, compared with echocardiography, may potentially be offset by reduced resource utilization. Although exploratory in nature, our observations raise the possibility that CMR prior to elective CABG coupled with targeted risk-reduction measures could lead to reduced resource utilization. This hypothesis is in line with contemporary healthcare recommendations [1], and merits prospective assessment.

## Abbreviations

• CABG

coronary artery bypass grafting

•

coronary artery disease

•
• CI

confidence interval

•
• CV

coefficient of variation

•
• LE

late enhancement

•
• LV

left ventricular

•
• LVEDV

LV end-diastolic volume

•
• LVEF

LV ejection fraction

•
• LVESV

LV end-systolic volume

•
• MI

myocardial infarction

•
• MR

magnetic resonance

•
• CMR

cardiac MR

•
• OR

odds ratio

•
• PCI

percutaneous coronary intervention

•
• PMO

persistent microvascular obstruction

•
• SI

signal intensity

•
• SV

stroke volume

This work was funded by the BHF (British Heart Foundation) Chair and Programme Grant BHF PG/02/128, and the Wellcome Trust Cardiovascular Functional Genomics Initiative 066780/2/012 to A.F.D. L.U.Z. was supported by the Swiss National Science Foundation (PBBSB-105860)/Lichtenstein-Stiftung Basel. C.D. was supported by a personal fellowship from the Deutsche Forschungsgemeinschaft (DE 826/1-1). The sponsors had no role in study design, data collection, data analysis, data interpretation or viewing of the study. We thank Mrs Margaret Kinnaird, and Dr Tom Martin, Dr Robin Weir and Dr Paddy Mark for guidance on CMR analyses.

## References

References
1
Gibbons

R. J.

Leading the elephant out of the corner: the future of health care. Presidential address at the American Heart Association 2006 Scientific Sessions
Circulation
2007
, vol.
115
(pg.
2221
-
2230
)
2
Ferguson

T. B.
Hammill

B. G.
Peterson

E. D.
DeLong

E. R.
Grover

F. L.

A decade of change: risk profiles and outcomes for isolated coronary artery bypass grafting procedures, 1990–1999. A report from the STS National Database Committee and the Duke Clinical Research Institute
Ann. Thorac. Surg.
2002
, vol.
73
(pg.
480
-
489
)
3
Rosamond

W.
Flegal

K.
Friday

G.
, et al
Heart disease and stroke statistics: 2007 update. A report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee
Circulation
2007
, vol.
115
(pg.
E69
-
E171
)
4
Eisenberg

M. J.
Filion

K. B.
Azoulay

A.
Brox

A. C.
Haider

S.
Pilote

L.

Outcomes and cost of coronary artery bypass graft surgery-in the United States and Canada
Arch. Int. Med.
2005
, vol.
165
(pg.
1506
-
1513
)
5
British Heart Foundation Heart Statistics
Treatment
2006
London
British Heart Foundation
(pg.
61
-
77
)
6
Pell

J. P.
Walsh

D.
Norrie

J.
, et al
Outcomes following coronary artery bypass grafting and percutaneous transluminal coronary angioplasty in the stent era: a prospective study of all 9890 consecutive patients operated on in Scotland over a two year period
Heart
2001
, vol.
85
(pg.
662
-
666
)
7
Lassnigg

A.
Hiesmayr

M. J.
Bauer

P.
Haisjackl

M.

Effect of centre-, patient- and procedure-related factors on intensive care resource utilisation after cardiac surgery
Intensive Care Med.
2002
, vol.
28
(pg.
1453
-
1461
)
8
Hein

O. V.
Birnbaum

J.
Wernecke

K.
England

M.
Konertz

W.
Spies

C.

Prolonged intensive care unit stay in cardiac surgery: Risk factors and long-termsurvival
Ann. Thorac. Surg.
2006
, vol.
81
(pg.
880
-
885
)
9
Higgins

T. L.

Quantifying risk and assessing outcome in cardiac surgery
J. Cardiothorac. Vasc. Anesth.
1998
, vol.
12
(pg.
330
-
340
)
10
Berry

C.
Murdoch

D. R.
McMurray

J.

Economics of chronic heart failure Eur
J. Heart Failure
2001
, vol.
3
(pg.
283
-
291
)
11
Tu

J. V.
Mazer

C. D.

Can clinicians predict ICU length of stay following cardiac surgery?
Can. J. Anaesth.
1996
, vol.
43
(pg.
789
-
794
)
12
Geissler

H. J.
Holzl

P.
Marohl

S.
, et al
Risk stratification in heart surgery: comparison of six score systems
Eur. J. Cardiothorac. Surg.
2000
, vol.
17
(pg.
400
-
405
)
13
Tu

J. V.
Mazer

C. D.
Levinton

C.
Armstrong

P. W.
Naylor

C. D.

A predictive index for length of stay in the intensive-care unit following cardiac-surgery
Can. Med. Assoc. J.
1994
, vol.
151
(pg.
177
-
185
)
14
Pennell

D. J.
Sechtem

U. P.
Higgins

C. B.
, et al
Clinical indications for cardiovascular magnetic resonance (CMR): Consensus Pane Report
Eur. Heart J.
2004
, vol.
25
(pg.
1940
-
1965
)
15
Schvartzman

P. R.
Srichai

M. B.
Grimm

R. A.
, et al
Nonstress delayed-enhancement magnetic resonance imaging of the myocardium predicts improvement of function after revascularization for chronic ischemic heart disease with left ventricular dysfunction
Am. Heart J.
2003
, vol.
146
(pg.
535
-
541
)
16
Selvanayagam

J. B.
Kardos

A.
Francis

J. M.
, et al
Value of delayed-enhancement cardiovascular magnetic resonance imaging in predicting myocardial viability after surgical revascularization
Circulation
2004
, vol.
110
(pg.
1535
-
1541
)
17
Zimmerli

L. U.
Mark

P. B.
Steedman

T.
, et al
Vascular function in patients with end-stage renal disease and/or coronary artery disease: a cardiac magnetic resonance imaging study
Kidney Int.
2007
, vol.
71
(pg.
68
-
73
)
18
Lorenz

C. H.
Walker

E. S.
Morgan

V. L.
Klein

S. S.
Graham

T. P.

Normal human right and left ventricular mass, systolic function, and gender differences by cine magnetic resonance imaging
J. Cardiovasc. Magn. Reson.
1999
, vol.
1
(pg.
7
-
21
)
19
Sievers

B.
Kirchberg

S.
Franken

U.
Puthenveettil

B. J.
Bakan

A.
Trappe

H. J.

Visual estimation versus quantitative assessment of left ventricular ejection fraction: a comparison by cardiovascular magnetic resonance imaging
Am. Heart J.
2005
, vol.
150
(pg.
737
-
742
)
20
Alfakih

K.
Plein

S.
Bloomer

T.
Jones

T.
Ridgway

J.
Sivananthan

M.

Comparison of right ventricular volume measurements between axial and short axis orientation using steady-state free precession magnetic resonance imaging
J. Cardiovasc. Magn. Reson.
2003
, vol.
18
(pg.
25
-
32
)
21
Mosteller

R. D.

Simplified calculation of body-surface area
N. Engl. J. Med.
1987
, vol.
317
pg.
1098

22
Simonetti

O. P.
Kim

R. J.
Fieno

D. S.
, et al
An improved MR imaging technique for the visualization of myocardial infarction
2001
, vol.
218
(pg.
215
-
223
)
23
Mark

P. B.
Johnston

N.
Groenning

B. A.
, et al
Redefinition of uremic cardiomyopathy by contrast-enhanced cardiac magnetic resonance imaging
Kidney Int.
2006
, vol.
69
(pg.
1839
-
1845
)
24
Baks

T.
van Geuns

R. J.
Biagini

E.
, et al
Recovery of left ventricular function after primary angioplasty for acute myocardial infarction
Eur. Heart J.
2005
, vol.
26
(pg.
1070
-
1077
)
25
Hombach

V.
Grebe

O.
Merkle

N.
, et al
Sequelae of acute myocardial infarction regarding cardiac structure and function and their prognostic significance as assessed by magnetic resonance imaging
Eur. Heart J.
2005
, vol.
26
(pg.
549
-
557
)
26
Lund

G. K.
Stork

A.
Saeed

M.
, et al
Acute myocardial infarction: evaluation with first-pass enhancement and delayed enhancement MR imaging compared with (TI)-T-201 SPECT imaging
2004
, vol.
232
(pg.
49
-
57
)
27
Fieno

D. S.
Kim

R. J.
Chen

E. L.
Lomasney

J. W.
Klocke

F. J.
Judd

R. M.

Contrast-enhanced magnetic resonance imaging of myocardium at risk – distinction between reversible and irreversible injury throughout infarct healing
J. Am. Coll. Cardiol.
2000
, vol.
36
(pg.
1985
-
1991
)
28
Berry

C.
Norrie

J. N.
Hogg

K.
Brett

M.
Stevenson

K.
McMurray

J. J. V.

The prevalence, nature and importance of hematologic abnormalities in heart failure
Am. Heart J.
2006
, vol.
151
(pg.
1513
-
1521
)
29
Bellenger

N. G.
Burgess

M. I.
Ray

S. G.
, et al
Comparison of left ventricular ejection fraction and volumes in heart failure by echocardiography, radionuclide ventriculography and cardiovascular magnetic resonance: are they interchangeable?
Eur. Heart J.
2000
, vol.
21
(pg.
1387
-
1396
)
30
Roques

F.
Nashef

S. A. M.
Michel

P.
, et al
Risk factors and outcome in European cardiac surgery: analysis of the EuroSCORE multinational database of 19030 patients
Eur. J. Cardiothorac. Surg.
1999
, vol.
15
(pg.
816
-
822
)
31
Higgins

T. L.
McGee

W. T.
Steingrub

J. S.
Rapoport

J.
Lemeshow

S.
Teres

D.

Early indicators of prolonged intensive care unit stay: Impact of illness severity, physician staffing, and pre-intensive care unit length of stay
Crit. Care Med.
2003
, vol.
31
(pg.
45
-
51
)
32
Nathanson

B.
Higgins

T. L.
Teres

D.
Copes

W. S.
Kramer

A.
Stark

M.

An updated method to assess ICU clinical performance and resource utilization
Crit. Care Med.
2005
, vol.
33
pg.
A26

33
Mounsey

J. P.
Griffith

M. J.
Heaviside

D. W.
Brown

A. H.
Reid

D. S.

Determinants of the length of stay in intensive-care and in-hospital after coronary-artery surgery
Br. Heart J.
1995
, vol.
73
(pg.
92
-
98
)
34
Michalopoulos

A.
Tzelepis

G.
Pavlides

G.
Kriaras

J.
Dafni

U.
Geroulanos

S.

Determinants of duration of ICU stay after coronary artery bypass graft surgery
Br. J. Anaesth.
1996
, vol.
77
(pg.
208
-
212
)
35
American Heart Association
Heart Disease and Stroke Statistics: 2003 Update
2002
Dallas
American Heart Association

## Author notes

1

These authors contributed equally to this study.