If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address for reprints: Naoki Yoshimura, MD, PhD, First Department of Surgery, University of Toyama, Graduate School of Medicine, 2630 Sugitani, Toyama, Toyama, 930-0194, Japan.
Translation Medical Center, National Center of Neurology and Psychiatry, Tokyo, JapanEndowed Course for Health System Innovation, Keio University School of Medicine, Tokyo, JapanDepartment of Healthcare Quality Assessment, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan
Department of Healthcare Quality Assessment, Graduate School of Medicine, The University of Tokyo, Tokyo, JapanDepartment of Health Policy and Management, Keio University School of Medicine, Tokyo, Japan
The present study developed a new risk model for congenital heart surgery in Japan and determined the relationship between hospital procedural volume and mortality using the developed model.
Methods
We analyzed 47,164 operations performed between 2013 and 2018 registered in the Japan Cardiovascular Surgery Database-Congenital and created a new risk model to predict the 90-day/in-hospital mortality using the Japanese congenital heart surgery mortality categories and patient characteristics. The observed/expected ratios of mortality were compared among 4 groups based on annual hospital procedural volume (group A [5539 procedures performed in 90 hospitals]: ≤50, group B [9322 procedures in 24 hospitals]: 51-100, group C [13,331 procedures in 21 hospitals]: 101-150, group D [18,972 procedures in 15 hospitals]: ≥151).
Results
The overall mortality rate was 2.64%. The new risk model using the surgical mortality category, age-weight categories, urgency, and preoperative mechanical ventilation and inotropic use achieved a c-index of 0.81. The observed/expected ratios based on the new risk model were 1.37 (95% confidence interval, 1.18-1.58), 1.21 (1.08-1.33), 1.04 (0.94-1.14), and 0.78 (0.71-0.86) in groups A, B, C, and D, respectively. In the per-procedure analysis, the observed/expected ratios of the Rastelli, coarctation complex repair, and arterial switch procedures in group A were all more than 3.0.
Conclusions
The risk-adjusted mortality rate for low-volume hospitals was high for not only high-risk but also medium-risk procedures. Although the overall mortality rate for congenital heart surgeries is low in Japan, the observed volume-mortality relationship suggests potential for improvement in surgical outcomes.
Graphical abstract
Graphical AbstractSummary of major findings of the present study are demonstrated. The bar height and error bars indicate the O/E ratio and its 95% CI estimated by bootstrap, respectively. JCVSD-Congenital, Congenital Section of Japan Cardiovascular Surgery Database; J-STAT,Japan Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Congenital Heart Surgery; O/E, observed/expected; TOF, tetralogy of Fallot; CoA, coarctation of the aorta; AVSD, atrioventricular septal defect.
A national database in Japan shows that congenital heart surgery in low-volume hospitals carries varying degrees of mortality risk in addition to procedural and patient-specific risks.
Congenital heart surgery in Japan has a low mortality rate, comparable to the STS/EACTS data, despite the small number of cases per hospital. However, whether or not a volume-outcome relationship exists among hospitals in Japan remains unclear. In this study, we identified several procedures that are prone to have a volume-mortality relationship, including some medium-risk procedures.
See Commentary on page 1551.
Despite recent advances in understanding the pathophysiology of complex cardiac anomalies, surgical techniques, and perioperative care, congenital heart surgery remains one of the most challenging procedures to perform in the entire field of surgery. Various quality improvement efforts have been made to enhance surgical outcomes.
Quality improvement in cardiovascular surgery: results of a surgical quality improvement programme using a nationwide clinical database and database-driven site visits in Japan.
In the past decades, the evaluation of patient outcomes has become increasingly accepted as an important step in assessing and improving the quality of patient care.
Japan Cardiovascular Surgery Database Organization Risk model of thoracic aortic surgery in 4707 cases from a nationwide single-race population through a web-based data entry system: the first report of 30-day and 30-day operative outcome risk models for thoracic aortic surgery.
Large multi-institutional database, such as The Society of Thoracic Surgeons (STS) database and the European Association for Cardiothoracic Surgery (EACTS) database, have developed and validated methods of risk adjustment in reported outcomes.
The Japanese Society for Cardiovascular Surgery launched the Congenital Section of the Japan Cardiovascular Surgery Database (JCVSD-Congenital) in 2008, which covers almost all congenital heart operations performed in Japan.
Current status of cardiovascular surgery in Japan 2013 and 2014: a report based on the Japan Cardiovascular Surgery Database. 2: congenital heart surgery.
Current status of cardiovascular surgery in Japan, 2013 and 2014: a report based on the Japan Cardiovascular Surgery Database (JCVSD). 1: mission and history of JCVSD.
Patients’ background characteristics and the combination of surgical procedures for the same disease are known to differ among Europe, the United States, and Japan,
and a classification for risk analysis unique to Japan has been sought. In 2020, we applied the same methodology as the STS-EACTS Congenital Heart Surgery (STAT) mortality categories/scores to the dataset of the JCVSD-Congenital and created the Japan STAT (J-STAT) mortality categories/scores,
Although the actual mortality rate depends on the characteristics of the patient as well as the type of surgery, there is still no accurate prediction model that incorporates both the J-STAT categories and the patient characteristics.
There are a large number of Japanese cardiac surgery programs, and on average, each program has a low volume compared with other countries.
Effect of procedural volume on outcome of coronary artery bypass graft surgery in Japan: implication toward public reporting and minimal volume standards.
In fact, most of the programs in Japan belong to the lowest volume category (<150 per year), and previous studies based on STS and EACTS data have shown that this is associated with poor early prognosis.
Surgical volume and center effects on early mortality after pediatric cardiac surgery: 25-year North American experience from a multi-institutional registry.
On the other hand, it has also been reported that the average mortality rates of congenital heart surgery in Japan was comparable to those in STS and EACTS data.
Current status of cardiovascular surgery in Japan 2013 and 2014: a report based on the Japan Cardiovascular Surgery Database. 2: congenital heart surgery.
Committee for Scientific AffairsThe Japanese Association for Thoracic Surgery Thoracic and cardiovascular surgeries in Japan during 2018: annual report by the Japanese Association for Thoracic Surgery.
Therefore, it remains unclear whether or not volume-mortality relationship exists in the Japanese healthcare system. The present study was performed to develop and validate a new risk model incorporating J-STAT categories and patient characteristics for congenital heart surgery in Japan and clarify the nationwide trend in hospital procedural volume and the association between the surgical volume and outcomes of congenital heart surgery using the new risk model.
Materials and Methods
This study was approved by the review board of the JCVSD-Congenital Section's Data Utilization Committee and subsequently received approval by the Institutional Review Board of Toyama University (R2021007) on April 7, 2021.
Study Population
The JCVSD-Congenital has developed a web-based data collection software system.
Patients undergoing any pediatric cardiac operation with or without cardiopulmonary bypass at Japanese hospitals participating in the JCVSD-Congenital between January 1, 2013, and December 31, 2018 (150 programs, 47,164 operations), were included in this study. We identified 186 procedures, defined by the JCVSD-Congenital. These 186 procedures were grouped into 5 categories (J-STAT categories), where category 1 has the lowest risk of mortality and category 5 the highest.
The most technically complex of all concurrent procedures in 1 operation was called the "primary procedure." If multiple operations were performed during the same hospitalization, only the primary procedure belonging to the highest J-STAT category at each hospitalization was analyzed.
Operative mortality was defined as the 90-day and in-hospital mortality, which included deaths within 90 days of operation, even when the patient had been discharged, as well as deaths during the same hospitalization as the operation even more than 90 days after the operation.
Patient Characteristics
The J-STAT category, age-weight categories at the time of surgery, sex, urgency, mechanical ventilation, and inotropic agents were used as explanatory variables, and operative mortality was used as the dependent variable. The definition of urgency was as follows: elective, the patient's cardiovascular status was stable in the days or weeks before the operation, so the procedure could have been deferred without an increased risk of a compromised outcome; urgent, the procedure was required during the same hospitalization to minimize the risk of further clinical deterioration; emergency, patients requiring emergency operations were likely to have ongoing severe cardiovascular compromise and were not responsive to any form of therapy except for cardiac surgery; and salvage, the patient was undergoing cardiopulmonary resuscitation en route to the operating room or before anesthesia induction or was receiving ongoing extracorporeal membrane oxygenation support to maintain their life.
Categorization of Hospitals Based on Procedural Volume
The hospital congenital heart surgery annual case volume was averaged over a 6-year period (2013-2018) and divided into 4 groups (group A: ≤50, group B: 51-100, group C: 101-150, group D: ≥151).
Statistical Analyses
Categorical variables were expressed as numbers and percentages. In descriptive statistics, cases with missing values for the items to be counted were excluded, and percentages were calculated using the number of cases with no missing values for the items as the denominator.
To evaluate the association between hospital volume and mortality adjusted for the effect of preoperative patient characteristics and surgical procedures on mortality risk, we built a model to predict the 90-day and in-hospital mortality using preoperative patient characteristics and surgical procedures and evaluated its predictive performance. In this analysis, cases with no missing values in the objective and explanatory variables were included. Data with all covariates (46,650 records: 98.9% of all record) were randomly divided into 2 subsets for model development, the training data set (37,321 records; 80%) and the test data set (9329 records; 20%). With the training data set, the current predictive model was built using a multivariable stepwise logistic regression model with the outcome variable being the 90-day and in-hospital mortality and the explanatory variables being the J-STAT category, sex, age, and weight categories at the time of surgery, urgency, mechanical ventilation, and inotropic agents.
The test data were then used to evaluate the generalization performance of the prediction model. The discriminative performance of the model was evaluated by the receiver operating characteristic curve and area under the curve (AUC), and the calibration was evaluated by Brier score.
whose explanatory variables were the J-STAT category, age, and weight categories at the time of surgery. The model that was superior in this comparison, but whose parameters (coefficients) were re-estimated using all of the data (ie, 46,650 records), was used in subsequent analyses.
The model with better performance, that is, the current predictive model as shown in the “Results” section, was used to estimate the expected number of deaths for each hospital volume stratum by summing the predicted probabilities for each stratum. The ratio of the expected number of deaths to the observed number of deaths (observed/expected [O/E]) and its 95% confidence interval (CI) were calculated using the bootstrap method. The number of excess deaths were calculated as the difference between the observed and the expected number of deaths. The number of transfers required to avoid 1 excess death was calculated as the number of patients divided by the number of excess deaths. All tests were 2 tailed. R version 4.0 or later
Among the 47,164 patients, 5552 (11.8%) were low-body-weight neonates, 5717 (12.1%) had nonelective surgeries, 2821 (6.0%) underwent mechanical ventilation, and 424 (0.9%) received inotropic agents preoperatively. The overall operative mortality rate was 2.64%. The operative mortality rates by J-STAT categories (category 1, 2, 3, 4, and 5) were 0.3%, 2.0%, 3.8%, 10.11%, and 15.62%, respectively (Table 1). Major complications were reoperation (2.8%), persistent neurologic deficit (0.8%), pacemaker (0.7%), paralyzed diaphragm (0.9%), chylothorax (3.1%), and mediastinitis (0.8%). Median postoperative length of stay was 16 days (interquartile range, 17 days).
Table 1Operative mortalities for each J-STAT mortality category and patients’ characteristics
n
Death
Mortality(%)
J-STAT mortality category
J-STAT 1
15,949
41
0.26
J-STAT 2
16,199
329
2.03
J-STAT 3
11,343
436
3.84
J-STAT 4
2463
249
10.11
J-STAT 5
1210
189
15.62
Sex
Female
22,583
580
2.6
Male
24,579
664
2.7
N/A
2
Age and weight categories
Age ≥1 y
21,529
271
1.26
Age: 1-11 mo, weight ≥6.0 kg
5446
52
0.95
Age: 1-11 mo, weight ≥4.0-5.9 kg
6662
118
1.77
Age: 1-11 mo, weight <4.0 kg
5085
276
5.42
Age <1 mo, weight ≥3.0 kg
2378
119
5.00
Age <1 mo, weight: 2.0-2.9 kg
3554
287
8.08
Age <1 mo, weight <2.0 kg
1998
114
5.71
N/A
512
Urgency
Elective
41,445
719
1.73
Urgent
4060
304
7.49
Emergency
1572
187
11.90
Salvage
85
34
40.00
N/A
2
Preoperative mechanical ventilation
No
44,343
902
2.03
Yes
2821
342
12.12
Preoperative inotropic agents
No
46,740
1206
2.58
Yes
424
38
8.96
J-STAT, Japan-Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Congenital Heart Surgery; N/A, not available.
In the new risk model, in addition to these 2 variables, sex, urgency, mechanical ventilation, and inotropic agents were also candidates for explanatory variables, and all except for sex were selected after variable selection was performed using the stepwise method. With the test data set, the AUC of each receiver operating characteristic curve improved significantly from 0.771 (95% CI, 0.745-0.797) in the previous model to 0.816 (95% CI, 0.791-0.840) in the current model (P < .001, Figure 1). Brier scores
for the test data were 0.0251 for the previous model and 0.0246 for the current model. Figure 2 shows the predicted and observed mortality rates according to the predicted risk, and the predictions are accurate from low to high mortality rates. Odds ratios and CIs of each patient characteristics for operative mortality as determined by a multivariable logistic regression analysis are shown in Table E1.
Figure 1Receiver operating characteristic (ROC) curves for the previous model (left) and the current model (right). The upper and lower curves indicate the range of 95% CI estimated by bootstrap.
Figure 2Predicted and observed mortality rates according to predicted risk. Cases were divided into 6 quantiles on the basis of the predicted mortality (Q1-Q6), and predicted and observed mortality rates were compared for each of the 6 risk groups. Q1, first quantile (predicted mortality: <0.2%); Q2, second quantile (predicted mortality: 0.2%-0.8%); Q3, third quantile (predicted mortality: 0.8%-1.2%); Q4, fourth quantile (predicted mortality: 1.2%-2.2%); Q5, fifth quantile (predicted mortality: 2.2%-4.3%); Q6, sixth quantile (predicted mortality: ≥4.3%).
Figure 3 summarizes the case distribution and mortality rates by hospital volume and J-STAT category. In total, 47,164 congenital heart surgeries were performed in 150 hospitals over 6 years. Fifteen (10%) hospitals performed 151 or more procedures per year, and 90 (60%) hospitals performed 50 or less procedures per year. These 47,164 operations were classified according to J-STAT categories (category 1: 15,949, category 2: 16,199, category 3: 11,343, and category 4 + 5: 3673). High-risk operations tended to be performed in high-volume centers: 1134 of 5539 (20.5%) operations performed in group A hospitals were categorized as 3 or greater, whereas 6539 of 18,972 (34.5%) operations in group D were categorized as 3 or greater. Only half of the hospitals in group A performed J-STAT 4 + 5 procedures. Even with the tendency to perform higher-risk operations in larger-volume hospitals, the unadjusted mortality rate tended to be lower among hospitals that performed more than 150 operations (3.0% in group A, 3.3% in group B, 2.8% in group C, and 2.1% in group D).
Figure 3Case distribution and mortality rates by annual hospital volume and J-STAT mortality category. J-STAT, Japan Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Congenital Heart Surgery.
The risk-adjusted expected number of deaths in groups A, B, C, and D were 118.8, 249.6, 359.8, and 508.8, respectively. The observed number of deaths in groups A, B, C, and D were 163, 301, 375, and 398, respectively. For the 90-day and in-hospital mortality, the O/E ratios in groups A, B, C, and D were 1.37 (95% CI, 1.18-1.58), 1.21 (1.08-1.33), 1.04 (0.94-1.14), and 0.78 (0.71-0.86), respectively (Figure 4 and Video Abstract). Thus, the overall O/E ratio in group A was approximately 1.8-fold higher than in group D.
Figure 4The O/E ratio of operative mortality by annual procedural volume group. The bar height and error bars indicate the O/E ratio and its 95% CI estimated by bootstrap, respectively. O/E, Observed/expected.
Figure 5 shows the O/E ratios in each group per J-STAT category. For the 90-day and in-hospital mortality in category 3 operations, the O/E ratios in group A, B, C, and D were 1.51 (1.15-1.90), 0.97 (0.79-1.18), 1.04 (0.87-1.21), and 0.88 (0.74-1.02), respectively. In category 4 + 5 operations, the O/E ratios in groups A, B, C, and D were 1.39 (0.91-1.89), 1.32 (1.11-1.54), 1.00 (0.84-1.16), and 0.83 (0.72-0.95), respectively. In category 3 and 4 + 5 operations, the O/E ratios in group A were approximately 1.7-fold higher than those in group D, respectively.
Figure 5The O/E ratio of operative mortality by annual procedural volume group and J-STAT mortality category. The bar height and error bars indicate the O/E ratio and its 95% CI estimated by bootstrap, respectively. J-STAT, Japan Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Congenital Heart Surgery; O/E, observed/expected.
The relationship between the O/E ratio and hospital procedural volume was evaluated for 15 common procedures (Figure 6 and Table E2). We identified several procedures that were more susceptible to the procedural volume-mortality relationship than others. The most prominent was the Rastelli procedure, with O/E ratios of 9.96 (2.46-18.13), 2.45 (0-5.42), 2.19 (0.44-4.27), and 0.62 (0.00-1.47) in groups A, B, C, and D, respectively. The second significant procedure was the arterial switch procedure, with O/E ratios of 8.83 (4.93-13.82), 2.67 (1.37-4.25), 2.31 (1.34-3.39), and 0.96 (0.47-1.52) in groups A, B, C, and D, respectively. The third significant procedure was coarctation complex repair, with O/E ratios of 6.39 (0-22.34), 0 (0-0), 0.89 (0-2.31), and 0.25 (0-0.80) in groups A, B, C, and D, respectively. Other procedures with a marked volume-mortality relationship included mitral valvuloplasty, tetralogy of Fallot repair, bidirectional Glenn procedure, and complete atrioventricular septal defect (AVSD) repair, all of which had O/E ratios of 3.00 or more in group A.
Figure 6The O/E ratios of operative mortality of common procedures by annual procedural volume group. The bar height and error bars indicate the O/E ratio and its 95% CI estimated by bootstrap, respectively. O/E, Observed/expected; ASD, atrial septal defect; VSD, ventricular septal defect; TOF, tetralogy of Fallot; CoA, coarctation of the aorta; PAB, pulmonary artery banding; AVSD, atrioventricular septal defect; SP shunt, systemic to pulmonary shunt; TAPVC, total anomalous pulmonary venous connection.
The number of excess deaths and the number of transfers required to avoid 1 excess death by hospital volume are shown in Table E3. The number of transfers required to avoid 1 excess death was less than 100 in patients less than 1 year or J-STAT 3 or higher in group A and less than 1 month or J-STAT 4 or higher in group B, suggesting that transferring these patients would be effective in efficiently reducing excess deaths.
Discussion
In the present study, we developed a highly discriminative model to predict mortality after congenital heart surgery using the J-STAT categories. As in the United States and Europe, a significant procedural volume-mortality relationship was observed in Japan. Furthermore, we identified several procedures that are particularly prone to have a volume-mortality relationship. The lowest-volume hospitals (≤50) had high adjusted mortality rates (O/E ratios >3) not only for high-risk procedures but also for medium-risk procedures, such as the Rastelli procedure and arterial switch procedure. These results suggest that the safety of an operation cannot be guaranteed when the number of cases per year is small (<50 cases per year).
New Risk Model for Congenital Heart Surgery
In the field of adult cardiac surgery, it is common to calculate risk factors based not only on the type of surgery and age but also on the preoperative conditions, such as hypertension, renal function, and the presence of diabetes, and adding these factors provides a high prediction accuracy.
However, in the field of pediatric cardiac surgery, the addition of such risk factors to the procedure did not necessarily provide good predictive accuracy, because there are many more types of procedures than in adult cardiac surgery; in addition, comorbidities like those seen in adults are rare, and having a comorbidity does not necessarily lead to an increased risk (eg, Down syndrome is protective for AVSD repair).
Contemporary outcomes of complete atrioventricular septal defect repair: analysis of the Society of Thoracic Surgeons Congenital Heart Surgery Database.
In the present study, we added preoperative status, which were urgency, age-weight categories, mechanical ventilation, and inotropic agents to the J-STAT complexity category, and this made the prediction more accurate than the previous model developed by Hirahara and colleagues
using only the procedural complexity category and age-weight categories. The AUC increased from 0.771 to 0.816, reflecting an increase in discriminating power.
There are 2 possible explanations for the prediction accuracy increasing by adding patient background as factors: (1) the need for emergency surgery, ventilatory management, and inotropic medication itself reflect a vulnerable patient background; and (2) these factors themselves contribute to worsening postoperative conditions, regardless of patient background. The improvement in accuracy can be attributed to both factors, not just one or the other.
The mortality predicted by our risk model and the observed mortality were comparable in both low- and high-risk groups, suggesting that our calibration was also valid. These results suggest that a risk model that incorporates patient characteristics into the J-STAT categories of procedural complexity can be a useful framework for calculating predicted mortality.
In the present study, the definition of operative mortality included 90-day and in-hospital mortality. Previous studies focused on the 30-day or in-hospital mortality regardless of the length of hospital stay.
Committee for Scientific AffairsThe Japanese Association for Thoracic Surgery Thoracic and cardiovascular surgeries in Japan during 2018: annual report by the Japanese Association for Thoracic Surgery.
Compared with such previous studies, we used longer-term outcome data, which may have contributed to more appropriate surveillance by avoiding missed deaths that can occur in later periods.
Effect of Procedural Volume on the Outcome of Congenital Heart Surgery
Although the caseload of hospitals in Japan was low compared with Europe and the United States, the average mortality rates of congenital heart surgery were comparable to those in STS and EACTS data.
Current status of cardiovascular surgery in Japan 2013 and 2014: a report based on the Japan Cardiovascular Surgery Database. 2: congenital heart surgery.
This has been the main argument against regionalization in Japan, but the exact reasons why the overall outcomes have been maintained despite the small number of cases are unknown. It may be due in part to the public health insurance system in Japan, which covers the majority of medical expenses and allows patients to receive expensive rescue treatments regardless of their financial status. Regionalization can lead to potentially better outcomes due to the accumulation of experience, more sustainable delivery of medical care, and more efficient use of healthcare resources but at the expense of healthcare access. The “optimal” caseload of a hospital ideally needs to be determined on a regional or national basis, depending on the volume-outcome relationship, geographic access to healthcare, and the type of national healthcare system in place.
Our study showed for the first time that the volume-outcome relationship of congenital heart disease surgery also holds in the Japanese healthcare system. The overall risk-adjusted mortality rate in the lowest-volume hospitals (≤50 per year) was 1.8 times higher than in the highest-volume hospitals. Kansy and colleagues
showed that the risk-adjusted in-hospital mortality was higher in low-volume (<150 per year) and medium-volume (150-250 per year) centers using the EACTS database (odds ratio referring to >350 cases: 1.45 and 1.84, respectively). Because the main volume range in our study corresponded to the lowest volume in that European study, this was the first study to show a significant volume-mortality relationship in this low-volume range. Welke and colleagues,
with STS data obtained between 2002 and 2006, showed a nonlinear volume-mortality relationship, with an inflection point at approximately 250 cases per year, below which the slope of the volume-mortality relationship became more pronounced. However, the analysis of Welke and colleagues
may have been influenced by the variability in the mortality in low-volume hospitals. The present study showed a more reliable volume-mortality relationship in the low-volume range by using more detailed categorization to eliminate the effect of variability among low-volume hospitals.
In the present study, we also identified several procedures that were particularly prone to have a volume-mortality relationship, including the Rastelli operation, coarctation complex repair, arterial switch operation, mitral valvuloplasty, tetralogy of Fallot repair, bidirectional Glenn procedure, and complete AVSD repair. Welke and colleagues
reported that more complex lesions have a higher sensitivity to the volume-mortality relationship, than less complex lesions. Surprisingly, in the present study, sensitive procedures included not only high-risk procedures but also medium-risk ones in terms of STAT/J-STAT categories.
This may be due to the smaller caseload than the previous study. Furthermore, it is notable that in group A, the O/E ratios for some procedures were 3 or more, whereas the overall O/E ratio was 1.37. Although the 95% CIs for the O/E ratios in the per-procedure analysis were wider because of the lower number, an O/E ratio of 3 or more is extremely high, considering that the O/E ratio for the arterial switch operation in the historically famous Bristol case was approximately twice as high as in other centers.
These results suggest that a small annual number of cases (≤50 per year) cannot guarantee safe surgery for congenital heart disease. However, it is difficult to distinguish whether the high O/E ratio in group A is due to the presence of a small number of even poorer-performing hospitals or whether it is a trend for all hospitals in group A. Even if these O/E ratios were assessed for each small-volume hospital, it would be difficult to obtain statistically meaningful results because the 95% CIs would likely be even wider. However, these summary data need to be shared with patients; with such high O/E ratios, patients may be willing to travel farther for surgery even at the expense of healthcare access.
In both the United States and Europe, regionalization has led to improved surgical outcomes, reflecting the volume-outcome relationship,
Surgical volume and center effects on early mortality after pediatric cardiac surgery: 25-year North American experience from a multi-institutional registry.
Regionalization is not only a matter of increasing the number of procedures but also a long, gradual process that requires the maturation of medical and co-medical teams and the gathering of healthcare resources.
Surgical volume and center effects on early mortality after pediatric cardiac surgery: 25-year North American experience from a multi-institutional registry.
In this study, we demonstrated a volume-outcome relationship at a low volume, suggesting that increasing the annual number of cases, even by 50, may lead to improved outcomes.
Study Limitations
There are several limitations in the present study. First, although multiple surgical procedures can be entered into the database, in this study, we conducted the analysis as 1 procedure per surgery. This is inevitable to avoid overly detailed categorization and a small number of cases in each category; however, as a result, we cannot deny the possibility that the combination of each procedure has been oversimplified and the risk subsequently underestimated. Second, the predictive model proposed in this study has only been internally validated and not externally validated. Because there are likely to be many fundamental differences in countries other than Japan, such as the number of cases per year and the education system of surgeons, caution should be exercised in interpreting the conclusions and extrapolating the results. Third, this study only shows an association between hospital volume and outcome, not a causal relationship. There would be confounding factors between hospital volume and outcomes that were not taken into account in this study, such as hospital type (eg, university hospitals, cardiovascular center hospitals, pediatric hospitals, and general hospitals), patient access, staffing of healthcare professionals, hospital resources, and collaboration with other departments within the hospital. Therefore, whether or not an increase in the number of hospital caseloads leads to better outcomes is a subject for future research. Finally, in this study, mortality was the only end point applied in examining the volume-outcome relationship. The morbidity, functional status, and neurologic status after surgery are also important outcomes that need to be elucidated in future investigations.
Conclusions
In Japan, the large number of hospitals for congenital heart surgeries has resulted in a low procedural volume per hospital. High-volume hospitals that perform 151 or more congenital heart surgeries per year had a significantly lower overall mortality after adjusting for the type of surgery and patient characteristics than the low-volume hospitals (≤100 per year). Furthermore, we identified several procedures that are particularly prone to have a volume-mortality relationship. The lowest-volume hospitals (≤50) had high adjusted mortality rates, not only for high-risk procedures but also for medium-risk procedures (Figure 7). These results will provide fundamental data supporting the debate on the pros and cons of regionalization of congenital heart surgery programs in Japan and may accelerate this process as well.
Figure 7Summary of major findings of the present study are demonstrated. The bar height and error bars indicate the O/E ratio and its 95% CI estimated by bootstrap, respectively. JCVSD-Congenital, Congenital Section of Japan Cardiovascular Surgery Database; J-STAT, Japan Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Congenital Heart Surgery; O/E, observed/expected; TOF, tetralogy of Fallot; CoA, coarctation of the aorta; AVSD, atrioventricular septal defect.
R.I. receives grants-in-aid for scientific research 20K08177. M.A. receives honoraria or payment for lectures, presentations, speakers bureaus, manuscript writing, or educational events from Japan Blood Products Organization, Teijin Pharma Limited. All other authors reported no conflicts of interest.
The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest.
The present study investigated the relationship between Japanese hospital procedural volume and mortality using a new risk model. We analyzed 47,164 operations registered in the Congenital edition of the Japan Cardiovascular Surgery Database and created a new risk model to predict the 90-day/in-hospital mortality using the following factors: J-STAT categories, age-weight categories, urgency, preoperative mechanical ventilation, and inotropic use. Comparisons among 4 groups stratified by annual hospital procedural volume with regard to O/E ratios of mortality were conducted. The overall O/E ratio for mortality in low-volume hospitals, defined as performing 50 or less procedures per year, was significantly higher than in higher-volume hospitals. We identified several procedures at these hospitals that were particularly prone to have a volume-mortality relationship, including the Rastelli operation, arterial switch operation, coarctation complex repair, and mitral valvuloplasty. Taken together, our data indicate that the lowest-volume hospitals had high adjusted mortality rates for both high- and medium-risk procedures. Video available at: https://www.jtcvs.org/article/S0022-5223(22)00703-6/fulltext.
Appendix E1
Table E1Odds ratios and confidence intervals of patient characteristics for operative mortality by multivariable logistic regression analysis
OR (95% CI)
P value
(Intercept)
0.00 (0.00-0.00)
<.001
J-STAT mortality category
1
—
2
5.28 (3.76-7.42)
<.001
3
9.81 (7.01-13.7)
<.001
4
19.8 (13.8-28.3)
<.001
5
34.9 (24.1-50.5)
<.001
Urgency
Elective
—
Urgent
2.00 (1.72-2.35)
<.001
Emergency
2.95 (2.44-3.57)
<.001
Salvage
13.9 (8.49-22.8)
<.001
Age-weight categories at the time of surgery
Age ≥1 y
—
Age 1-11 mo, weight ≥6.0 kg
0.64 (0.47-0.87)
.005
Age 1-11 mo, weight 4.0-5.9 kg
0.99 (0.78-1.24)
.90
Age 1-11 mo, weight <4.0 kg
1.74 (1.44-2.09)
<.001
Age <1 mo, weight ≥3.0 kg
0.99 (0.78-1.27)
.95
Age < 1 mo, weight 2.0-2.9 kg
1.44 (1.18-1.76)
<.001
Age < 1 mo, weight <2.0 kg
1.30 (1.01-1.69)
.044
Preoperative mechanical ventilation
No
—
Yes
2.57 (2.21-2.99)
<.001
Preoperative inotropic agents
No
—
Yes
1.80 (1.25-2.60)
.001
OR, Odds ratio; CI, confidence interval; J-STAT, Japan Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Congenital Heart Surgery.
Table E3Excess deaths by hospital volume and number of transfers needed to avoid 1 excess death
A: 1-50
B: 51-100
C: 101-150
D: 151-
No. of excess deaths
Age
<1 mo
8.1
20.1
0.8
−29.0
1-5 mo
15.3
−3.7
5.9
−45.8
6-11 mo
7.5
11.4
4.5
4.9
≥1 y
13.3
23.6
4.0
−40.9
J-STAT mortality category
J-STAT 1
1.3
6.1
−1.6
−5.8
J-STAT 2
16.6
20.2
12.2
−49.0
J-STAT 3
18.9
−2.4
4.8
−21.4
J-STAT 4 + 5
7.3
27.4
-0.1
−34.6
No. of transfers necessary to avoid 1 excess death
Age
<1 mo
128.0
74.9
2895.0
−103.2
1-5 mo
77.8
−611.6
530.2
−95.8
6-11 mo
85.6
104.7
409.6
527.0
≥1 y
189.6
176.8
1467.3
−218.5
J-STAT
J-STAT 1
1762.1
526.0
−2817.2
−953.4
J-STAT 2
114.8
148.9
352.3
−139.8
J-STAT 3
50.3
−963.1
693.4
−221.1
J-STAT 4 + 5
23.3
24.1
−7359.9
−51.8
The number of patients transferred to avoid 1 excess death less than 100 is highlighted in bold. J-STAT, Japan Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery Congenital Heart Surgery.
Quality improvement in cardiovascular surgery: results of a surgical quality improvement programme using a nationwide clinical database and database-driven site visits in Japan.
Japan Cardiovascular Surgery Database Organization
Risk model of thoracic aortic surgery in 4707 cases from a nationwide single-race population through a web-based data entry system: the first report of 30-day and 30-day operative outcome risk models for thoracic aortic surgery.
Current status of cardiovascular surgery in Japan 2013 and 2014: a report based on the Japan Cardiovascular Surgery Database. 2: congenital heart surgery.
Current status of cardiovascular surgery in Japan, 2013 and 2014: a report based on the Japan Cardiovascular Surgery Database (JCVSD). 1: mission and history of JCVSD.
Effect of procedural volume on outcome of coronary artery bypass graft surgery in Japan: implication toward public reporting and minimal volume standards.
Surgical volume and center effects on early mortality after pediatric cardiac surgery: 25-year North American experience from a multi-institutional registry.
Contemporary outcomes of complete atrioventricular septal defect repair: analysis of the Society of Thoracic Surgeons Congenital Heart Surgery Database.
In the current issue of the Journal, Yoshimura and colleagues1 from Japan analyzed 47,164 congenital heart surgeries (CHS) performed between 2013 and 2018 from the Japan Cardiovascular Surgery Database. They compared observed/expected mortality among 150 programs grouped based on the annual hospital procedural volume: group A (≤50 cases, n = 90), group B (51-100 cases, n = 24), group C (101-150 cases, n = 21), and group D (>150 cases, n = 15). For comparison, they used a Japanese risk model that they termed J-STAT, which is very similar to the STAT model applied in North America.
00703-6&id=fx1.jpg){kind=link}
00703-6&id=fx2.jpg){kind=link}
00703-6&id=gr1.jpg){kind=link}
00703-6&id=gr2.jpg){kind=link}
00703-6&id=gr3.jpg){kind=link}
00703-6&id=gr4.jpg){kind=link}
00703-6&id=gr5.jpg){kind=link}
00703-6&id=gr6.jpg){kind=link}
00703-6&id=gr7.jpg){kind=link}