| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Cardiac Surgery, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
b Department of Healthcare Quality Assessment, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
Accepted for publication August 1, 2008.
* Address correspondence to Dr Motomura, Department of Cardiac Surgery, Faculty of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-8655, Japan (Email: noboru{at}motomura.org).
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsCommentReferences |
|---|
Methods: We analyzed 7133 CABG-only records from 97 participating sites throughout Japan using a data entry form with 255 variables that was sent to the JACVSD office by our Web-based data collection system. The statistical model was constructed by multiple logistic regression. Model discrimination was tested using the area under the receiver operating characteristic curve (C index). Model calibration was tested by the Hosmer-Lemeshow test.
Results: Of 7133 operations, 47.2% had diabetes mellitus, 14.0% were urgent, and 15.6% involved peripheral vascular disease. The observed 30-day and operative mortality rates were 2.02% and 2.72%, respectively. Significant variables with high odds ratios included emergency or salvage status (3.71), preoperative creatinine value exceeding 3.0 mg/dL (3.59), aortic valve stenosis (3.01), and moderate to severe chronic lung disease (2.86). Hosmer-Lemeshow test and C-index values for 30-day mortality were satisfactory at 0.96 and 0.85, respectively.
Conclusions: The results obtained in Japan were at least as good as those reported elsewhere. The performance of our risk model also matched those of the Society of Thoracic Surgeons National Adult Cardiac Database and the European Society Database.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
The number of CABG procedures has increased steadily in Japan, where cardiovascular disease was regarded as being less common than in Western countries. The Japanese lifestyle is now closer to those of Europe and the United States, especially the daily diet; however, a number of differences still remain. Genetics, the degree of mixing in racial backgrounds, education, workload, access to medical facilities, and social systems vary considerably. The purposes of this large-scale benchmark project with an integrated database are to evaluate the surgical procedures used and to improve the quality of cardiac surgery.
To assess daily practice in Japan, we need to build our own database system rather than borrow from other systems such as the STS NCD or the EuroSCORE system. We started a project to set up a nationwide integrated database in the cardiac surgery field, the Japan Adult Cardiovascular Surgery Database (JACVSD). The structure and content of the JACVSD are similar to those of the STS NCD. This is our first report of the risk model of isolated CABG in an East Asian population, which is comparable with such models used in Western countries and suggests new approaches for improving the quality of cardiac surgery.
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
The JACVSD Database
The JACVSD was inaugurated in 2000 to assess surgical outcomes after cardiovascular surgical procedures on a multicenter basis. The database currently captures clinical information from 151 hospitals as of July 2007 (28.5% of all units performing CABG in 2005), and we included in this study 111 hospitals that had enrolled by December 2005.
The data collection form has 255 variables that are almost identical to those of the STS NCD (available online at http://sts.org). The definitions of JACVSD variables (available online at http://www.jacvsd.umin.jp) are the same as those of the STS NCD. JACVSD created software for a Web-based data collection system, through which each data manager in each participating hospital sends data.
Although participation in the JACVSD is voluntary, data completeness is high. The overall preoperative risk factors were missing in fewer than 2% of the whole of the assembled data. The accuracy of submitted data was maintained by data auditing, which involves a number of administrative office members visiting a randomly chosen hospital on a monthly basis and checking the data against the clinical records. The ratio of JACVSD data registered to the true number of cases at the hospital has further been confirmed by comparing the data reported in advance to the Japanese Association for Thoracic Surgery (JATS) Registry [1]. We excluded 14 centers from 111 hospitals because they entered less than 90% of the amount of JATS Registry data. The data from 97 hospitals were analyzed. Exclusion of the 14 centers did not affect the preoperative risk for the outcome.
Study Population
We examined only isolated CABG operations between January 1, 2000, and December 31, 2005, excluding those in which valve or other major surgical procedures were involved. From the set of records with the CABG-only procedural designation, those in which the patient's age was outside the range selected and those in which sex was missing were excluded from this analysis. About 5% of records were missing 30-day mortality and were also excluded. Finally, after this data standardization process, the population for this risk model analysis comprised 7133 CABG-only patients from 97 participating sites throughout Japan, and their records were accessed.
Auditing
To perform the audit on a routine basis, we created the Site Visit Working Group (SV-WG). The WG members consist of one SV-WG chief (selected from the administrative office members) and 6 data managers from six areas in Japan. One hospital is randomly picked every month, and the SV-WG chief lists all the deceased patients and in advance makes printed tables that show all the entered variables of all the deceased patients. The chief also makes another table that includes the cases randomly picked up from the nondeceased patients.
The SV-WG (4 to 6 members) visits the selected hospital, and one member verifies the number of procedures from the original operative record list in the hospital and identifies it with the number of the JACVSD data. Other members look up the clinical charts of all the deceased patients and compare them with the printed table. The audit takes a day, from 9 A.M. to 5 P.M. If extra hours are available, members check the nondeceased patients using their clinical records and the tables that the SV-WG chief has prepared.
Since this audit system started as a routine practice from 2006, 21 hospitals have been audited so far. In the near future the SV-WG members will perform the audit in the area in their charge, and such audits can then be done in several hospitals at the same time every month all over Japan.
End Points
The primary outcome measure of JACVSD was 30-day mortality, defined as death within 30 days of an operation, regardless of the patient's geographic location, even if the patient had been discharged from the hospital. The second outcome measure of analysis was the operative mortality, defined as in-hospital or 30-day mortality (whichever was longer), which is equivalent to "the 30-day operative mortality" as defined in the STS NCD. Major morbidity was defined as any of five postoperative in-hospital complications: stroke, reoperation for any reason, need for mechanical ventilation for more than 24 hours postoperatively, renal failure with newly required dialysis, or deep sternal wound infection [2]. In this analysis we use composite 30-day mortality or major morbidity as the third end point.
Statistical Analysis
The statistical model was multiple logistic regression. The variables entered in the model were selected using bivariate tests, 
2 tests for categoric covariates, and unpaired t tests or Wilcoxon rank sum tests for continuous covariates. All variables significant at the p < 0.2 level were entered into the model provided they were present in at least 2% of the sample. Nonsignificant variables were eliminated from the model one at a time, beginning with the variable having the highest p value. The stability of the model was checked every time a variable was eliminated. In the case of continuous variables where the relationship with outcome was not linear, such as preoperative creatinine levels, we determined cutoff points.
When all statistically nonsignificant variables had been eliminated from the model, goodness-of-fit testing was used to assess how good the discrimination was between survivors and nonsurvivors, and the area under the receiver operating characteristic (ROC) curve was used to assess how well the model could discriminate between patients who lived from patients who died. Model calibration (the degree to which observed outcomes are similar to the predicted outcomes from the model, compared across patient groups) was examined by comparing observed average with predicted average within each of 10 equally sized subgroups arranged in increasing order of patient risk. To evaluate model calibration, the Hosmer-Lemeshow test for the lack of goodness of fit was applied [3].
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
|
|
|
demonstrates the calibration of models; that is, how well the predicted event rates match the observed event rates among patient subgroups of risk. The details of model performance metrics are compiled in Table 4.
|
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
It is easy to calculate the expected mortality of our patients preoperatively by using the EuroSCORE or STS NCD calculator [4], but the results would not be based on actual data obtained from our own patient population, and it will not help to improve our quality control. In fact, Yap and colleagues [5] reported that the additive and logistic EuroSCORE did not accurately predict the outcomes of the Australian CABG population and that the use of the EuroSCORE models for risk prediction may not be appropriate in their patients.
It is for these reasons that we built our own database. Of course, it is very useful to examine these two large databases and compare them with our database. But it is more important to create our own database to understand our own situation and problems. Another reason was uncontrolled broadcasting by the mass media on surgical outcomes based on their questionnaire, which had no scientific background. Many newspapers and magazines have provided feature articles on hospital rankings of procedure volume. These data had no statistical background or validation, and such activities are futile [6]. Therefore, we realized that this is the time to create a scientific database ourselves and then to publish outcomes and findings specific to the population in Japan.
The characteristics of the studied patients in our database were similar to those reported by the STS NCD. The distribution of men, peripheral vascular disease, and the number of diseased vessels were almost identical to their counterparts in STS. In our Japanese study, there was less chronic lung disease (5.5% vs 15.0%, p < 0.0001 by 2 test), more elective cases (78.2% vs 61.7%, p < 0.0001), more cerebrovascular disease (14.5% vs 10.5%, p < 0.0001), and more DM patients (47.2% vs 31.6%) than those obtained in the STS.
The almost threefold difference in chronic lung disease was larger than the differences in the other variables. The smoking rate in Japan was much higher in the past, but has declined recently, so that the current rate of Japanese smokers in this study did not show as great a difference as the STS NCD rate (21.5% vs 18.1%). We do not know the exact reason for this great difference in chronic lung disease. Whether the population of isolated CABG patients was different from or similar to that in STS and other major databases, our own original database is needed because the outcome of medical services is affected by nonmedical environments.
The rate of DM in the patient population was 47.2%, which was a much higher rate than in Western countries and much higher than we expected. In the STS NCD, the ratio of DM was 31.6% and the OR of DM on operative mortality was 1.15. In contrast, DM was not included in a risk factor for 30-day mortality and operative mortality in this study. In our current version of database, a question about DM was not focused specifically on insulin treatment. Our next database version (version 4), which we are working on, will have a variable regarding the insulin treatment. Then, being a DM patient with insulin treatment would be a risk factor in the next version. In fact, DM was a risk factor of composite 30-day mortality and major morbidity even in this version (Table 3).
Table 3 presents the ORs of this risk model. These results were in general similar to those of the STS NCD. Reoperation was a more serious variable in the United States, and aortic stenosis was more serious in Japan. Also, the OR for renal dysfunction was higher in Japan than in the United States: a creatinine level of more than 3.0 mg/dL had an OR of 3.59 in our study, whereas that for renal failure and dialysis in the STS NCD was 1.88. The general condition of patients with renal failure and dialysis is worse in Japan than in the United States, because the health system in Japan is generous enough for such patients to survive longer. In addition, patients with better condition were to be treated by catheter interventions and only far more serious patients were referred to CABG. As a result, an older dialysis population with more severely complicated disease has been sent to cardiac surgeons in Japan.
The C indexes were 0.85 for 30-day mortality and 0.86 for operative mortality. These values were much higher than those in other databases [7, 8]. Generally, the smaller the number of entries that are found in the data set, the better the value of the C index. The Hosmer-Lemeshow test is also affected by the sample number. We used only 7133 records in this study—much fewer than the number used in the STS NCD—but the reliability of our database was still satisfactory. Because of the relatively small sample size, we did not divide these samples into two parts, one for analysis and the other for validation. We are accumulating sample data daily through our Web-based data entry system, and we will update our analysis using a much larger sample size in the near future to perform an updated analysis.
The operative mortality of this study in 2005 was 2.72%, which was a result similar to or better than other major nationwide database reports [9, 10]. Shroyer and colleagues [2] reported that the operative mortality of isolated CABG in 1997 to 1999 was 3.05% according to the STS NCD analysis. However, the latest report from the STS Database Web site showed a significant decrease in the operative mortality of unadjusted isolated CABG from 3.1% in 1997 to 2.1% in 2006 (http://www.sts.org/sections/stsnationaldatabase/publications/executive/article.html), which was better than our data of 2005. Composite major mortality or morbidity showed similar results: 13.64% vs 13.40%. However, Shroyer and colleagues' [2] data were collected several years before ours. The STS NCD Committee reported that mortality has improved over the last 10 years, even though surgeons had to face older, sicker, and higher-risk patients [11]. So their mortality rate might be better than ours if it were updated. In any event, these data indicate that the outcomes of cardiac interventions in Japan were as good as those of the United States.
The Japanese Association for Thoracic Surgery (JATS) has been maintaining a registry of cardiovascular procedures in Japan since 1986, and the latest version of the JATS Annual Report in 2004 showed that the number of cardiovascular units and isolated CABG in Japan was 539 and 19,930, respectively [1]. An average of 37.0 isolated CABG procedures were performed per year in all institutions. The total number of acquired operations (CABG + valve + thoracic aorta + arrhythmia) was 42,781, which was an average of 79.4 per year at each hospital. Some units are doing only pediatric congenital cardiac operations, and this average is greatly underestimated. But the number is definitely much smaller than in other countries.
The Japanese Board of Cardiovascular Surgery indicates that Japan had 1911 Board-certified cardiac surgeons as of March 2007 [12] who treated an average of 18.7 cases per year, including valve and arrhythmia procedures. Again, this number is much lower than it is in other countries. Nevertheless, the 30-day mortality in Japan is not worse than in those countries. The reason for the excellent results achieved by Japanese cardiac surgeons appears to be multifactorial [13]. The health insurance system in Japan covers most medical expenses, and a small individual payment is made at each operation [14]. Patients can remain in the hospital beyond the first week or two, if needed, without falling victim to the burden of excessive extra charges, and they can choose any surgeon and any hospital that they prefer, regardless of their insurance coverage. Surgeons can perform their surgical and medical duties at the maximum level with considerably less pressure from the hospital or the insurance company that would be felt in other countries.
Because of the small case volume, Japanese surgeons are able to hold both preoperative and postoperative conferences on each patient more intensively than is possible abroad. Moreover, usually, cardiac operations are not performed by a single consultant but rather by one or two consultant surgeons with trainees. Because a surgeon's salary is kept at a fixed, flat rate in Japan, each surgeon must take responsibility not only for himself but also for other surgeons. Surgeons usually do not have to operate twice a day in one operating theater, and they can spend ample time on both planning and accomplishing all surgical procedures with a high level of competence.
Postoperative care in an intensive care unit or cardiac care unit is mainly the task of doctors who are trainee surgeons or intensive care specialists, and not by the nursing staff, a system that may offer better care than the other systems [15]. Postoperative intensive treatment should prolong the patient's life beyond 30 days. Under these socioeconomic circumstances, surgical patients in Japan are able to receive the full benefit of the highest class of medical services [13].
As time periods are defined, 30-day mortality is usually shorter than operative mortality. In our series the difference between these two values was larger than STS NCD (2.02% and 2.72% in Japan vs 1.7% and 2.0% in STS NCD). The intensive postoperative treatment in Japan described above could explain the large discrepancy between 30-day mortality and operative mortality.
Theoretically, quality control can be influenced not only by surgical procedures but also by the nature of the processes used in surgical practice [16]. It is currently common practice to assess such processes during procedures. Our database also has such variables, for example, internal mammary artery usage, preoperative β-blocker usage, and so on. Although we could not include the process measurement in this article, we are planning in our next step to examine our data from the viewpoint of process measurement.
The final goal of a clinical database is to improve the quality of our daily practice. The STS NCD proved this effect in the early stage and published in 2002 [17]. But they estimated between 1990 and 1999, which took 10 years to prove the preferable effect even by their large database. Our database has just started, and another several years are needed to show the positive effect of the database on our daily practice.
In conclusion, this is our first report of the risk model of isolated CABG in the Japanese population using a newly developed database system. The patient profile was slightly different from those in the large databases in Western countries, but the risk-adjusted mortality and risk factors were similar. Although patient backgrounds and socioeconomic conditions differ, surgical outcomes in this field were achieved to the same level of excellence as in Western countries. Further analysis is needed, using our database, in the fields of valvular and aortic surgery. These analyses will contribute to maintaining a high level of quality in the Japanese heart surgeon's daily practice.
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentReferences |
|---|
Related Article
This article has been cited by other articles:
|
|
 |
|
  H. Miyata, N. Motomura, H. Tsukihara, S. Takamoto, and Japan Cardiovascular Surgery Database Risk models including high-risk cardiovascular procedures: clinical predictors of mortality and morbidity Eur J Cardiothorac Surg, May 1, 2011; 39(5): 667 - 674. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Saito, N. Motomura, H. Miyata, S. Takamoto, S. Kyo, M. Ono, and Japan Cardiovascular Surgery Database Organization Age-specific risk stratification in 13488 isolated coronary artery bypass grafting procedures Interact CardioVasc Thorac Surg, April 1, 2011; 12(4): 575 - 581. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. H. Edwards Invited Commentary Ann. Thorac. Surg., December 1, 2008; 86(6): 1872 - 1872. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
