Several studies have recorded an association between surgeon and hospital procedure volume and the outcome of various interventions,1,2 including coronary artery bypass grafting (CABG).1–4 The major clinical practice guidelines recommend minimum procedure volumes, both for surgeons and for hospitals,5,6 and while these recommendations do not exceed evidence level C, they have important implications for health care planning.7,8 However, the evidence for an association between procedure volume and CABG outcome sits alongside reports of good results at hospitals with a low procedure volume.9,10 Furthermore, the association between volume and mortality appears to hold up only to a volume threshold, beyond which further volume increases yield no further improvements.4 In the United States, the evidence for an association between volume and mortality has weakened as a result of the obligatory publication of CABG mortality data.11 In Spain, the relationship between CABG volume and mortality is not firmly established. Gutacker et al. reported notably higher CABG mortality in Spain (5%) than in England (2.3%) and attributed this difference to differences in procedure volume (43 hospitals with 154 CABG procedures per year in Spain vs 29 hospitals with 690 procedures per year in England).12 In a more recent study of CABG surgery in 5 European countries, the same authors established a “safe threshold” minimum volume of 415 CABG procedures per center per year.4 However, although mean CABG per center in Spain falls short of this target,13 risk-adjusted CABG mortality in Spain (EuroSCORE scale14) is lower than predicted.13

The goal of the present study was to examine the association between CABG volume and outcome variables (in-hospital mortality in the index episode, 30-day readmission, and mortality during 30-day readmissions) in hospitals of the Spanish National Health System (NHS).

The number of index CABG events selected was 17 335 (figure 1), and these events were associated with an unadjusted mortality rate of 5.0% and a median hospital stay of 13 [9-20] days. Mean patient age was 67.8±9.7 years, and 78.7% of patients were men. Of the selected index events, 11 151 (64.3%) were isolated CABG procedures, and this subset was associated with a mean patient age of 65.8±9.7 years and a higher proportion of men (83.1%). In-hospital mortality for isolated CABG surgery was 3%, and the median hospital stay was 12 [8-19] days. Total and isolated CABG showed statistically significant differences in patient age, unadjusted in-hospital mortality, and mean hospital stay (P <.001).

Figure 1.

Selection of index events. AMI, acute myocardial infarction; CABG, coronary artery bypass grafting; MDC, major diagnostic category; PIN, personal identification number.

(0.14MB).

The most frequent comorbidities in CABG patients are shown in table 1. For all CABG patients, the 30-day readmission rate for cardiac causes was 8.6%, and the mortality associated with these readmissions was 5%. The corresponding figures for isolated CABG were 8.3% and 3%, respectively. The most frequent principal diagnoses triggering readmission are listed in the table 1 of the supplementary data.

The area below the ROC curve for adjusted in-hospital mortality was 0.87 (for both total and isolated CABG). Factors identified as significant for risk adjustment are shown in the supplementary data, together with their corresponding models and discrimination capacity, which was good for index-event and 30-day mortality but poor for 30-day readmission (table 2, table 3, table 4, and table 5 of the supplementary data and figure 1, figure 2, and figure 3 of the supplementary data).

A marked and statistically significant correlation was noted between index-event RAMR after complex CABG (coinciding with major cardiac surgery) and after isolated CABG (r=0.843; P <.001) (figure 2). Index-event and 30-day RAMR showed no significant correlation with the volume of complex or isolated CABG (r=–0.04; P=.8 in both cases); in contrast, RARR was significantly correlated to both total CABG (r=–0.318; P=.029) and isolated CABG (r=–0.306; P=.036).

Figure 2.

Linear correlation between RAMR for isolated CABG and for complex CABG (CABG coinciding with other major cardiac surgery). CABG, coronary artery bypass grafting; RAMR, risk-adjusted in-hospital mortality rate.

(0.1MB).

The k-means clustering algorithm identified high-volume hospitals as those with an annual mean of ≥ 155 CABG procedures over the study period; for the analysis of isolated CABG, high-volume hospitals were those averaging ≥ 95 isolated CABG procedures per year. In total, 33 centers (70%) performed fewer than 155 CABG procedures per year, and only 13% complied with the European Society of Cardiology (ESC) guideline recommendation to achieve a mean annual rate of ≥ 200 CABG procedures.6 Events treated at high-volume centers were associated with a lower RAMR (5.81±2.07 vs 4.96±1.56; P <.001) (figure 3). Cardiac-related 30-day RARR and associated 30-day RAMR were consistently lower at high-volume centers, for both total and isolated CABG (table 2).

Figure 3.

RAMR for events treated in high-volume and low-volume hospitals. In the left panel, high volume (≥ 155 CABG/year) and low volume (< 155 CABG/year) are defined by the k-means clustering algorithm. In the right panel, high volume (≥ 200 CABG/year) and low volume (< 200 CABG/year) are as recommended in the European Society of Cardiology clinical practice guidelines on myocardial revascularization.6 In both comparisons, RAMR is significantly higher in low-volume hospitals (17% in the k-means clustering comparison and 18% according to the ESC guideline recommendation). CABG, coronary artery bypass grafting; RAMR, risk-adjusted in-hospital mortality rate.

(0.19MB).

High-volume hospitals did not have a higher proportion of isolated CABG procedures (64.6% vs 64.1% in low-volume hospitals; P=.474) but did have a higher proportion of elective admissions for CABG (68.7% vs 66.3%: P < .001). The association of risk factors with in-hospital mortality after elective CABG (total) is compared between low-volume and high-volume centers in table 3. Low-volume centers had a higher prevalence of comorbidities overall, but differences in age and sex were not significant. The same pattern was observed for isolated CABG (table 6 of the supplementary data).

The main findings of our study are as follows. a) The volume of CABG in the Spanish NHS is low, both per center and per year, and only 13% of hospitals meet the European Society of Cardiology clinical practice guideline recommendation to perform ≥ 200 procedures per year.6b) Procedure volume at NHS centers shows a consistent and significant association with outcome (mortality and readmission) for total and isolated GAGB, with adjusted in-hospital mortality 17% higher in low-volume centers. c) The association between high GAGB volume and better outcome is likely related to the higher rate of complications in low-volume centers. d) Centers achieving better outcomes with isolated CABG tend also to have better results with complex CABG.

Compared with health care systems in other countries, the Spanish NHS achieves a low mean number of CABG procedures per hospital,4,9–12 and therefore the definitions of high and low volume in the current study denote, in relative terms, 2 low-volume CABG categories. The low CABG volume in NHS hospitals is probably related to the low mean surgeon volume and is below guideline recommended rates.6

Unlike Gutacker et al.,4,12 we found major differences in outcome between low-volume and high-volume NHS hospitals, both for total and isolated CABG. The correlation between RAMR for complex CABG and isolated CABG indicates that the performances of each center in both procedures are interrelated. However, this result is also compatible with the failure-to-rescue concept,22 which posits that the capacity to effectively treat the main severe postsurgery complications depends more on the general operation of the hospital than on the surgical procedure itself. An influence of failure to rescue in the present study is suggested by the higher rate of comorbidities associated with CABG at low-volume centers. These comorbidities included acute myocardial infarction, stroke, pneumonia, and—for isolated CABG—cardiogenic shock. All of these comorbidities influence mortality risk adjustment; however, they are highly unlikely to have been present on admission for elective CABG and are thus likely to be postsurgery complications. The higher frequency of these comorbidities in low-volume centers forces the adjustment to operate in the opposite direction to the hypothesis, thus likely masking larger differences between high-volume and low-volume centers.

The association between structural variables and outcomes is complex. Because procedure volume per se does little more than ensure reasonable confidence intervals around statistical estimates, Luft8 has argued that policies to exclude or close centers according to their activity are less effective than alternative policies based on systematic information about risk-adjusted outcomes.23 Indeed, volume criteria cannot provide even a rough guide to policy so long as the NHS does not have access to robust and publicly available risk-adjusted outcome indicators supported by consensus between scientific societies and health care authorities.24 The publication of regularly updated, appropriately risk-adjusted outcomes is thus essential to allow analysis of the true effect of procedure volume and other factors. Toward this goal, our findings lead us to recommend a) greater concentration of CABG in high-volume centers within the Spanish NHS7,25 and b) the publication of outcomes.

Hospitals in the Spanish NHS have a low mean volume of CAGB procedures. We found an association between higher volume and improved outcome for CABG. The findings of this study indicate that CABG should be concentrated in high-volume centers within the Spanish NHS and support the publication of risk-adjusted outcomes of coronary intervention.

This study was supported by an unconditional award from the Fundación Interhospitalaria de Investigación Cardiovascular.

None declared.