Low-volume coronary artery bypass surgery: Measuring and optimizing performance

Abstract

See related article on page 1306. See related article on page 1306. In the current issue of the Journal, Miyata and colleagues1Miyata H. Motomura N. Ueda Y. Matsuda H. Takamoto S. Effect of procedural volume on outcome of CABG surgery in Japan: implication toward public reporting and minimal volume standards.J Thorac Cardiovasc Surg. 2008; 135: 1306-1312Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar describe the relationship between coronary artery bypass grafting (CABG) procedural volume and outcome in Japan. In reality, however, there are no high-volume programs in this study. What the authors have actually provided us is the most extensive study of low-volume and extremely low-volume CABG surgery in the literature. It complements previous studies from the United States that include some programs with low volumes, and it provides a striking counterpoint to New York studies that are weighted toward the high end of the volume spectrum. This report illustrates the potential for good performance at low volumes, as well as the statistical challenge of accurately measuring performance when sample sizes are small. It raises a number of unresolved issues in the ongoing volume–outcome debate, at least as applied to CABG surgery. For example, some payers and other stakeholders continue to promote best-practice volume requirements that are increasingly beyond the grasp of many programs, particularly as overall CABG volumes decrease nationally. Is this appropriate policy given the available outcomes data? Because many lower-volume programs function at a high level, can the public be protected while at the same time not penalizing such excellent programs? Is there a rational lower volume limit for CABG surgery programs? Are there better ways to measure performance that are less compromised by small sample sizes? Are there specific process and structural approaches that might promote optimal functioning of small programs? CABG is unique: it is a mature, standardized procedure that is performed more frequently than any other complex operation and that has also been scrutinized more thoroughly. Notwithstanding the general validity of the volume–outcome relationship for a number of medical conditions and surgical procedures, data from a variety of sources suggest that many low-volume CABG providers achieve excellent results. We believe the fundamental issue with low-volume CABG surgery is not inherently poor performance but rather the difficulty in accurately measuring performance. These 2 perspectives have quite different implications. If it were clear that low-volume CABG providers were uniformly poor performers, immutably limited by their lack of sufficient “practice,” then the only reasonable solution would be volume thresholds. For some highly complex but very infrequently performed procedures, such as esophagectomy or pancreatectomy, this might well be a justifiable approach. In reality, however, excellent performance is achieved by many CABG providers whose volumes, although they might number in the hundreds annually, do not meet the thresholds of organizations like the Leapfrog Group (450 procedures per year). In this circumstance, volume standards would unfairly stigmatize or penalize such high-quality but low-volume providers. Furthermore, at a time when CABG volumes are decreasing, such thresholds could also have unintended negative consequences. Given the importance of cardiac surgery to most institutions, failure to meet guidelines for “center of excellence” status or premium reimbursement might have substantial adverse implications. This could result in a perverse incentive to relax appropriateness criteria to meet volume thresholds, which might have a net negative effect on the health system. If one views low-volume CABG providers as a heterogeneous group, many of whom provide excellent results, then the main issue is how to accurately measure the performance of individual programs, a challenge with small sample sizes and limited mortality events. More comprehensive and robust approaches to performance measurement could be developed that are less limited by such concerns, and specific programmatic initiatives could also be implemented to facilitate high performance in smaller programs. We will examine both the evidence for a CABG volume–outcome relationship as well as statistical problems with assessing performance in low-volume programs. Findings from the study of Miyata and colleagues1Miyata H. Motomura N. Ueda Y. Matsuda H. Takamoto S. Effect of procedural volume on outcome of CABG surgery in Japan: implication toward public reporting and minimal volume standards.J Thorac Cardiovasc Surg. 2008; 135: 1306-1312Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar will be reviewed in the context of these two issues. Finally, recommendations will be presented to enhance both performance and its measurement in low-volume CABG programs. Although the strength of the CABG volume–outcome relationship is probably exaggerated in some studies by failure to account for sample size and clustering (eg, through the use of hierarchical models),2Urbach D.R. Austin P.C. Conventional models overestimate the statistical significance of volume-outcome associations, compared with multilevel models.J Clin Epidemiol. 2005; 58: 391-400Abstract Full Text Full Text PDF PubMed Scopus (67) Google Scholar there is little question that some association exists. This was evident in the original work of Luft and associates3Luft H.S. Bunker J.P. Enthoven A.C. Should operations be regionalized? The empirical relation between surgical volume and mortality.N Engl J Med. 1979; 301: 1364-1369Crossref PubMed Scopus (1366) Google Scholar nearly 30 years ago, and it has been demonstrated in numerous subsequent studies, including those from the modern era.4Birkmeyer J.D. Siewers A.E. Finlayson E.V.A. Stukel T.A. Lucas F.L. Batista I. et al.Hospital volume and surgical mortality in the United States.N Engl J Med. 2002; 346: 1128-1137Crossref PubMed Scopus (3910) Google Scholar, 5Shahian D.M. Normand S.L. The volume-outcome relationship: from Luft to Leapfrog.Ann Thorac Surg. 2003; 75: 1048-1058Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar, 6Welke K.F. Barnett M.J. Vaughan Sarrazin M.S. Rosenthal G.E. Limitations of hospital volume as a measure of quality of care for coronary artery bypass graft surgery.Ann Thorac Surg. 2005; 80: 2114-2119Abstract Full Text Full Text PDF PubMed Scopus (31) Google Scholar, 7Rathore S.S. Epstein A.J. Volpp K.G. Krumholz H.M. Hospital coronary artery bypass graft surgery volume and patient mortality, 1998-2000.Ann Surg. 2004; 239: 110-117Crossref PubMed Scopus (89) Google Scholar, 8Hannan E.L. Kilburn Jr., H. Bernard H. O'Donnell J.F. Lukacik G. Shields E.P. Coronary artery bypass surgery: the relationship between inhospital mortality rate and surgical volume after controlling for clinical risk factors.Med Care. 1991; 29: 1094-1107Crossref PubMed Scopus (244) Google Scholar, 9Hannan E.L. The relation between volume and outcome in health care.N Engl J Med. 1999; 340: 1677-1679Crossref PubMed Scopus (125) Google Scholar, 10Wu C. Hannan E.L. Ryan T.J. Bennett E. Culliford A.T. Gold J.P. et al.Is the impact of hospital and surgeon volumes on the in-hospital mortality rate for coronary artery bypass graft surgery limited to patients at high risk?.Circulation. 2004; 110: 784-789Crossref PubMed Scopus (56) Google Scholar, 11Hannan E.L. Wu C. Ryan T.J. Bennett E. Culliford A.T. Gold J.P. et al.Do hospitals and surgeons with higher coronary artery bypass graft surgery volumes still have lower risk-adjusted mortality rates?.Circulation. 2003; 108: 795-801Crossref PubMed Scopus (169) Google Scholar The strongest data supporting a volume–outcome association come from New York, although these studies include very few programs that are truly low volume, and their findings might not be generalizable. In 2004, for example, there were 39 New York programs providing isolated CABG surgery, and 75% of these programs had volumes of greater than 214 procedures. It remains uncertain whether the CABG volume–outcome relationship applies to all patients10Wu C. Hannan E.L. Ryan T.J. Bennett E. Culliford A.T. Gold J.P. et al.Is the impact of hospital and surgeon volumes on the in-hospital mortality rate for coronary artery bypass graft surgery limited to patients at high risk?.Circulation. 2004; 110: 784-789Crossref PubMed Scopus (56) Google Scholar or primarily to those at higher risk.12Nallamothu B.K. Saint S. Ramsey S.D. Hofer T.P. Vijan S. Eagle K.A. The role of hospital volume in coronary artery bypass grafting: is more always better?.J Am Coll Cardiol. 2001; 38: 1923-1930Abstract Full Text Full Text PDF PubMed Scopus (89) Google Scholar, 13Peterson E.D. Coombs L.P. DeLong E.R. Haan C.K. Ferguson T.B. Procedural volume as a marker of quality for CABG surgery.JAMA. 2004; 291: 195-201Crossref PubMed Scopus (248) Google Scholar Notably, the volume strata and mortality ranges for CABG are quantitatively unique among complex procedures in which the volume–outcome association has been investigated. In studies by Birkmeyer and associates4Birkmeyer J.D. Siewers A.E. Finlayson E.V.A. Stukel T.A. Lucas F.L. Batista I. et al.Hospital volume and surgical mortality in the United States.N Engl J Med. 2002; 346: 1128-1137Crossref PubMed Scopus (3910) Google Scholar using claims data on 901,667 Medicare patients, low and high volume hospital categories for CABG were < 230 procedures and > 849 procedures respectively, with an absolute mortality difference of only 1.1% (4.5% vs 5.6%) between these two extremes. In contrast, the low and high volume ranges were <1 and >16 procedures for pancreatecomy and <2 and >19 procedures for esophagectomy, with absolute differences in adjusted mortality that were orders of magnitude greater (16.3% vs 3.8% for pancreatectomy, 20.3% vs 8.4% for esophagectomy) than those for CABG. Perhaps because it is a mature and frequently performed procedure, the volume–outcome association for CABG is weak. Studies by Peterson and coworkers13Peterson E.D. Coombs L.P. DeLong E.R. Haan C.K. Ferguson T.B. Procedural volume as a marker of quality for CABG surgery.JAMA. 2004; 291: 195-201Crossref PubMed Scopus (248) Google Scholar using 2000–2001 data from the Society of Thoracic Surgeons (STS) National Adult Cardiac Database, adjusted for risk factors and clustering, demonstrated only a 0.07% decrease in mortality for every additional 100 procedures (P = .004). Because there was substantial variability in mortality in all strata of volume, there was limited ability to discriminate among providers based solely on volume. Similar findings were noted in a study of 228,738 patients by Rathore and associates7Rathore S.S. Epstein A.J. Volpp K.G. Krumholz H.M. Hospital coronary artery bypass graft surgery volume and patient mortality, 1998-2000.Ann Surg. 2004; 239: 110-117Crossref PubMed Scopus (89) Google Scholar using data from the 1998–2000 Nationwide Inpatient Sample. In both studies, the vast majority of low-volume providers are distributed widely and symmetrically about the mean, with variation increasing at progressively smaller program volumes. Scatterplots of observed mortality versus volume in these studies look strikingly similar to the funnel appearance of the 95% confidence intervals of a binomial event, with an average occurrence rate of about 2% to 3%, taken at various sample sizes.14Spiegelhalter D.J. Funnel plots for comparing institutional performance.Stat Med. 2005; 24: 1185-1202Crossref PubMed Scopus (544) Google Scholar This is illustrated in Figure 1, a scatterplot based on the 2004 isolated CABG results from Massachusetts, New York, Ontario, and California. Superimposed scatterplot smoothers are roughly horizontal, showing little volume–outcome association. Much of the variability in mortality at low volumes, regarded by many as an indicator of inconsistent performance, is quite likely explained by sampling error. An alternative approach to volume thresholds is outcomes profiling. Public reporting of CABG outcomes is favored by many policymakers and has been mandated by law in states like New York and Massachusetts. Properly performed (by no means a trivial caveat), such reports are reasonably objective, they provides transparency and accountability, and they address the most important interest of patients: operative survival. However, they are the most demanding in terms of the need for high-quality data, audit and validation, and appropriate analytic methodologies. Even with larger sample sizes, comparative assessment of provider performance can be challenging, especially when based on a single outcome such as mortality. This becomes increasingly problematic as sample sizes (program volumes) decrease, a feature illustrated previously with regard to volume–outcome studies. Outcomes profiling generates estimates of provider performance derived from a snapshot in time, typically a year of clinical activity. Such observed results are used to estimate “true” underlying program quality, ideally with confidence intervals that indicate how certain we are about this point estimate. As noted previously, the statistical confidence intervals around point estimates of mortality, an infrequent binomial event, become quite wide with small sample sizes (annual program volumes).5Shahian D.M. Normand S.L. The volume-outcome relationship: from Luft to Leapfrog.Ann Thorac Surg. 2003; 75: 1048-1058Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar Much of the variation of annual mortality rates among low-volume programs (eg, fewer than 100–150 procedures per year), as shown in Figure 1, can be largely explained by random statistical fluctuation, and this in turn limits the ability to draw firm conclusions about program quality.5Shahian D.M. Normand S.L. The volume-outcome relationship: from Luft to Leapfrog.Ann Thorac Surg. 2003; 75: 1048-1058Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar, 15Dimick J.B. Welch H.G. Birkmeyer J.D. Surgical mortality as an indicator of hospital quality: the problem with small sample size.JAMA. 2004; 292: 847-851Crossref PubMed Scopus (381) Google Scholar In studies by Dimick and colleagues,15Dimick J.B. Welch H.G. Birkmeyer J.D. Surgical mortality as an indicator of hospital quality: the problem with small sample size.JAMA. 2004; 292: 847-851Crossref PubMed Scopus (381) Google Scholar CABG was the only complex procedure performed with sufficient frequency by most programs to detect a doubling of mortality rate based on 3-year aggregate data. However, as CABG mortality rates continue to decrease, the sample sizes necessary to detect meaningful differences increase correspondingly. This brings us to the study by Miyata and colleagues1Miyata H. Motomura N. Ueda Y. Matsuda H. Takamoto S. Effect of procedural volume on outcome of CABG surgery in Japan: implication toward public reporting and minimal volume standards.J Thorac Cardiovasc Surg. 2008; 135: 1306-1312Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar in the current issue of the Journal, an extreme and revealing illustration of both the “problems” of low-volume surgery, as well as some potential solutions. The authors describe the demographics of CABG programs in Japan, where annual volumes are so uniformly low that there is simply no US analog.3Luft H.S. Bunker J.P. Enthoven A.C. Should operations be regionalized? The empirical relation between surgical volume and mortality.N Engl J Med. 1979; 301: 1364-1369Crossref PubMed Scopus (1366) Google Scholar, 4Birkmeyer J.D. Siewers A.E. Finlayson E.V.A. Stukel T.A. Lucas F.L. Batista I. et al.Hospital volume and surgical mortality in the United States.N Engl J Med. 2002; 346: 1128-1137Crossref PubMed Scopus (3910) Google Scholar, 7Rathore S.S. Epstein A.J. Volpp K.G. Krumholz H.M. Hospital coronary artery bypass graft surgery volume and patient mortality, 1998-2000.Ann Surg. 2004; 239: 110-117Crossref PubMed Scopus (89) Google Scholar In Japan there has been limited regulatory oversight of cardiac surgery program proliferation or performance prior to the past few years. Relative to their population size, lower incidence of coronary disease, and number of isolated CABG procedures, the number of CABG providers in Japan far exceeds that of any publicly reported US states or the province of Ontario, as demonstrated in Table 1. Overall, 76% of Japanese programs perform fewer than 50 CABG procedures annually. Only 5.6% of 540 Japanese cardiac surgery programs (representing most of the programs in the country) performed at least 100 CABG procedures annually between January 2001 and December 2004, and 24.6% of programs performed fewer than 15 procedures annually. Median annual CABG volume was 28 procedures per year during this period (interquartile range, 15–49 procedures), and the average annual volumes ranged from 0.25 to 293 isolated CABG procedures. Using a threshold of 150 procedures annually, 98.3% of Japanese programs would be classified as low or very low volume by US standards. Based on the findings of Miyata and colleagues,1Miyata H. Motomura N. Ueda Y. Matsuda H. Takamoto S. Effect of procedural volume on outcome of CABG surgery in Japan: implication toward public reporting and minimal volume standards.J Thorac Cardiovasc Surg. 2008; 135: 1306-1312Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar 94% of Japanese CABG programs would fall to the left of the vertical line (annual volume of 100 procedures) in Figure 1, which would be very low volume by US and Canadian standards.Table 12004 CABG volumes, numbers of programs, and outcomes for selected states and provinces compared with estimated annual Japanese experienceRegionPopulation aged ≥18 y∗For the US States: US Census Bureau Population Estimates 2006 (http://www.census.gov/popest/states/asrh/SC-EST2006-01.html); for Ontario: Statistics Canada, Canada's National Statistics Agency, Age and Sex for the Population of Canada, Provinces, Territories, Census Metropolitan Areas and Census Agglomerations, 2001 and 2006 Censuses (http://www12.statcan.ca/english/census06/data/topics/RetrieveProductTable.cfm?ALEVEL=3&APATH=3&CATNO=&DETAIL=0&DIM=&DS=99&FL=0&FREE=0&GAL=0&GC=99&GK=NA&GRP=1&IPS=&METH=0&ORDER=1&PID=88984&PTYPE=88971&RL=0&S=1&ShowAll=No&StartRow=1&SUB=0&Temporal=2006&Theme=66&VID=0&VNAMEE=&VNAMEF=&GID=837983); for Japan: Statistics Bureau, 2005 Population Census, First Basic Complete Tabulation (http://www.stat.go.jp/English/data/kokusei/2005/kihon1/00/hyodai.htm).2004 isolated CABG admissions2004 CABG admissions per 100,000 adults2004 CABG programs2004 CABG programs per 1000 CABG admissions2004 isolated CABG volume, median (min-max)2004 crude operative†Operative mortality was defined as follows: for California, Massachusetts, Ontario, and Pennsylvania, all deaths occurring within 30 days of surgical intervention, regardless of where the patient died; for New Jersey, all deaths up to 30 days after surgical intervention or deaths occurring during the hospital stay in which the operation was performed, regardless of the number of days after the procedure; for New York, all deaths within the hospitalization, all discharges (alive or dead) to hospice care except those still alive at 30 days, and all other 30-day deaths. mortality (%)2004 RAMR (%), median (min-max)‡New Jersey and Pennsylvania present risk-adjusted mortality rates graphically and do not report specific numbers.California§California CABG Outcomes Reporting Program (2007). Coronary Artery Bypass Graft Surgery in California: 2003–2004 Hospital and Surgeon Data. CA Office of Statewide Planning and Development; http://www.oshpd.ca.gov/HID/Products/Clinical_Data/CABG/03-04fullreport.pdf.26,924,93519,10170.91206.3120 (4–975)3.293.30 (0–12.5)New Jersey‖New Jersey Department of Health and Senior Services (May 2007). Cardiac Surgery in New Jersey 2004. Office of the Commissioner; http://www.nj.gov/health/healthcarequality/documents/cardconsumer04.pdf.6,635,222617793.1172.7290 (102–755)1.98NAMassachusetts¶Massachusetts Data Analysis Center (October 2006). Adult Coronary Artery Bypass Graft Surgery in the Commonwealth of Massachusetts: January 1, 2004–December 31, 2004. Department of Health Care Policy, Harvard Medical School; http://www.massdac.org/reports/CABG%202004.pdf.4,988,309398679.9143.5287 (101–537)2.012.09 (1.50–3.95)New York#New York State Department of Public Health (June 2006). Adult Cardiac Surgery in New York State: 2002–2004; http://www.health.state.ny.us/diseases/cardiovascular/heart_disease/docs/cabg_2002-2004.pdf.14,791,84112,98887.8393.0288 (1–1188)2.092.01 (0–3.6)OntarioGuru V, Wang J, Donovan L, Tu JV (June 2006). Report on Coronary Artery Bypass Surgery in Ontario: Fiscal Years 2002–2004; http://www.ccn.on.ca/memberpdfs/Report-CAB-Surg-Ontario-2002-2004.pdf.9,439,990719676.2111.5647 (305–945)1.241.11 (0.46–4.11)PennsylvaniaPennsylvania Health Care Cost Containment Council (February 2006). Pennsylvania's Guide to Coronary Artery Bypass Graft Surgery 2004; http://www.phc4.org/reports/cabg/04/docs/cabg2004report.pdf.9,635,74813,359138.6604.5195 (56–888)2.31NAJapan (current study)Miyata and colleagues (current study).1 2004 volume estimated from 2001–2004 aggregate data. Number of programs (540) represents, per Miyata and colleagues, almost all CABG programs in Japan. Crude mortality rate is based on 2001–2004 aggregate data. Risk-adjusted results are not presented because they are based on only a small sample of overall programs.105,943,70720,000 (Miyata and colleagues (current study).1 2004 volume estimated from 2001–2004 aggregate data. Number of programs (540) represents, per Miyata and colleagues, almost all CABG programs in Japan. Crude mortality rate is based on 2001–2004 aggregate data. Risk-adjusted results are not presented because they are based on only a small sample of overall programs.)18.9540+2728 (0.25–293) (Miyata and colleagues (current study).1 2004 volume estimated from 2001–2004 aggregate data. Number of programs (540) represents, per Miyata and colleagues, almost all CABG programs in Japan. Crude mortality rate is based on 2001–2004 aggregate data. Risk-adjusted results are not presented because they are based on only a small sample of overall programs.)1.92 (Miyata and colleagues (current study).1 2004 volume estimated from 2001–2004 aggregate data. Number of programs (540) represents, per Miyata and colleagues, almost all CABG programs in Japan. Crude mortality rate is based on 2001–2004 aggregate data. Risk-adjusted results are not presented because they are based on only a small sample of overall programs.)NACABG, Coronary artery bypass grafting; RAMR, risk-adjusted mortality rate; NA, not applicable.∗ For the US States: US Census Bureau Population Estimates 2006 (http://www.census.gov/popest/states/asrh/SC-EST2006-01.html); for Ontario: Statistics Canada, Canada's National Statistics Agency, Age and Sex for the Population of Canada, Provinces, Territories, Census Metropolitan Areas and Census Agglomerations, 2001 and 2006 Censuses (http://www12.statcan.ca/english/census06/data/topics/RetrieveProductTable.cfm?ALEVEL=3&APATH=3&CATNO=&DETAIL=0&DIM=&DS=99&FL=0&FREE=0&GAL=0&GC=99&GK=NA&GRP=1&IPS=&METH=0&ORDER=1&PID=88984&PTYPE=88971&RL=0&S=1&ShowAll=No&StartRow=1&SUB=0&Temporal=2006&Theme=66&VID=0&VNAMEE=&VNAMEF=&GID=837983); for Japan: Statistics Bureau, 2005 Population Census, First Basic Complete Tabulation (http://www.stat.go.jp/English/data/kokusei/2005/kihon1/00/hyodai.htm).† Operative mortality was defined as follows: for California, Massachusetts, Ontario, and Pennsylvania, all deaths occurring within 30 days of surgical intervention, regardless of where the patient died; for New Jersey, all deaths up to 30 days after surgical intervention or deaths occurring during the hospital stay in which the operation was performed, regardless of the number of days after the procedure; for New York, all deaths within the hospitalization, all discharges (alive or dead) to hospice care except those still alive at 30 days, and all other 30-day deaths.‡ New Jersey and Pennsylvania present risk-adjusted mortality rates graphically and do not report specific numbers.§ California CABG Outcomes Reporting Program (2007). Coronary Artery Bypass Graft Surgery in California: 2003–2004 Hospital and Surgeon Data. CA Office of Statewide Planning and Development; http://www.oshpd.ca.gov/HID/Products/Clinical_Data/CABG/03-04fullreport.pdf.‖ New Jersey Department of Health and Senior Services (May 2007). Cardiac Surgery in New Jersey 2004. Office of the Commissioner; http://www.nj.gov/health/healthcarequality/documents/cardconsumer04.pdf.¶ Massachusetts Data Analysis Center (October 2006). Adult Coronary Artery Bypass Graft Surgery in the Commonwealth of Massachusetts: January 1, 2004–December 31, 2004. Department of Health Care Policy, Harvard Medical School; http://www.massdac.org/reports/CABG%202004.pdf.# New York State Department of Public Health (June 2006). Adult Cardiac Surgery in New York State: 2002–2004; http://www.health.state.ny.us/diseases/cardiovascular/heart_disease/docs/cabg_2002-2004.pdf.∗∗ Guru V, Wang J, Donovan L, Tu JV (June 2006). Report on Coronary Artery Bypass Surgery in Ontario: Fiscal Years 2002–2004; http://www.ccn.on.ca/memberpdfs/Report-CAB-Surg-Ontario-2002-2004.pdf.†† Pennsylvania Health Care Cost Containment Council (February 2006). Pennsylvania's Guide to Coronary Artery Bypass Graft Surgery 2004; http://www.phc4.org/reports/cabg/04/docs/cabg2004report.pdf.‡‡ Miyata and colleagues (current study).1Miyata H. Motomura N. Ueda Y. Matsuda H. Takamoto S. Effect of procedural volume on outcome of CABG surgery in Japan: implication toward public reporting and minimal volume standards.J Thorac Cardiovasc Surg. 2008; 135: 1306-1312Abstract Full Text Full Text PDF PubMed Scopus (44) Google Scholar 2004 volume estimated from 2001–2004 aggregate data. Number of programs (540) represents, per Miyata and colleagues, almost all CABG programs in Japan. Crude mortality rate is based on 2001–2004 aggregate data. Risk-adjusted results are not presented because they are based on only a small sample of overall programs. Open table in a new tab CABG, Coronary artery bypass grafting; RAMR, risk-adjusted mortality rate; NA, not applicable. By comparison, during 2000–2001, the median volume of CABG procedures among STS National Adult Cardiac Database participants was 253,13Peterson E.D. Coombs L.P. DeLong E.R. Haan C.K. Ferguson T.B. Procedural volume as a marker of quality for CABG surgery.JAMA. 2004; 291: 195-201Crossref PubMed Scopus (248) Google Scholar notably still less than the Leapfrog threshold of 450 procedures. At the high-volume extreme, New York has had a longstanding aggressive approach to monitoring and improving cardiac surgery quality.16Hannan E.L. Kilburn Jr., H. Racz M. Shields E. Chassin M.R. Improving the outcomes of coronary artery bypass surgery in New York State.JAMA. 1994; 271: 761-766Crossref PubMed Scopus (658) Google Scholar Between 1997 and 1999, median CABG volume at New York hospitals was 527 procedures (mean, 577 procedures; interquartile range, 331–816 procedures). Only 2.14% of patients undergoing CABG were treated at hospitals performing fewer than 200 such procedures annually,10Wu C. Hannan E.L. Ryan T.J. Bennett E. Culliford A.T. Gold J.P. et al.Is the impact of hospital and surgeon volumes on the in-hospital mortality rate for coronary artery bypass graft surgery limited to patients at high risk?.Circulation. 2004; 110: 784-789Crossref PubMed Scopus (56) Google Scholar, 11Hannan E.L. Wu C. Ryan T.J. Bennett E. Culliford A.T. Gold J.P. et al.Do hospitals and surgeons with higher coronary artery bypass graft surgery volumes still have lower risk-adjusted mortality rates?.Circulation. 2003; 108: 795-801Crossref PubMed Scopus (169) Google Scholar and only about one tenth of New York hospitals had annual CABG volumes of less than 200 procedures. Given the consistently low volume of most Japanese CABG programs, their overall results will come as a surprise to many. The most complete data source for this study was a survey of 540 programs collected by the Japanese Association for Thoracic Surgery, including almost all programs in the country. The overall mortality rate was 1.9%, and mortality for all volume categories above 41 to 50 procedures per year was less than 2%, which is comparable with rates in most US state and national CABG registries. Mortality rates for programs with annual volumes of less than 41 to 50 procedures ranged from 2.42% to 3.15%. Because these voluntary survey data lacked both adequate risk adjustment and careful audit, the authors also studied a small subset of 36 highly selected Japanese programs that contributed data to the Japanese Adult Cardiovascular Surgery Database, a clinical registry modeled after the STS National Adult Cardiac Database. Table 2 of their article demonstrates that patients in these programs had a distribution of risk factors not dissimilar to what would be observed in many US CABG registries. Unadjusted 30-day mortality was 1.88%, and operative mortality (including in-hospital deaths occurring after 30 days) was 2.55%. Risk-adjusted 30-day mortality was 1.50% for programs with an annual volume of 51 or more procedures and 2.14% for hospital volumes of 31 to 50 procedures. These aggregate data illustrate the feasibility of achieving good overall performance at low volumes, but they do not address the problem of accurately measuring individual hospital performance based on small samples. Although individual hospital volumes and outcomes are not provided by the authors, one would presume that at median volumes of 28 procedures pe