Impact of certification status of the institute and surgeon on short-term outcomes after surgery for thoracic esophageal cancer: evaluation using data on 16,752 patients from the National Clinical Database in Japan

Esophagectomy for thoracic esophageal cancer continues to have high morbidity and mortality rates today [1]. It, therefore, requires a surgeon with experience in the relevant surgical techniques and an institute with a well-trained medical staff providing perioperative care. In 2009, the Japan Esophageal Society (JES) established a certification system for surgeons and institutes to contribute to improving national medical care by enhancing the professional knowledge and skills of esophageal surgeons. A qualified surgeon or institute would be certified as a “Board Certified Esophageal Surgeon” (BCES) or “Authorized Institute for Board Certified Esophageal Surgeon” (AIBCES). The first certification of BCESs and AIBCESs by the JES was in 2010 and 2013, respectively. As of Jan 1, 2019, there were 280 BCESs and 156 AIBCESs nationwide. Using data from the National Database of Hospital-based Cancer Registries in Japan, we recently verified the appropriateness of the certification system for AIBCES by the JES by examining the survival outcomes among patients with thoracic esophageal cancer who underwent an esophagectomy at an AIBCES or a Non-AIBCES [2]. The results revealed that 5-year survival outcomes were better for patients treated at an AIBCES than at a Non-AIBCES. In that report, however, we did not address an important aspect of surgical outcome: morbidity and mortality. In the present study, we analyzed data from the Japanese National Clinical Database (NCD) to compare short-term outcomes between AIBCES and Non-AIBCES. In addition, using a hierarchical multivariable logistic regression analysis, we determined which certification has a greater impact on surgery-related mortality, AIBCES, or BCES.

Enrolled in the study were 16,752 patients operated on at 854 institutes by 1879 surgeons. The annual average numbers of patients at AIBCESs and Non-AIBCESs were 27.8 and 3.7, respectively. The annual average numbers of patients treated by BCESs and Non-BCESs were 13.3 and 3.5, respectively.

This study produced several noteworthy results. First, 29% of patients received esophagectomy for thoracic esophageal cancer at a Non-AIBCES by a Non-BCES in Japan. Second, patients with more severe conditions were more frequently operated on at Non-AIBCSs than AIBCES and more frequently by Non-BCESs than BCESs. Third, hierarchical multivariable logistic regression analysis of the incidence of surgery-related mortality adjusted for patient-level risk factors revealed that the surgery-related mortality rate was significantly lower in the AIBCES/Non-BCES (OR: 0.54) or AIBCES/BCES (OR: 0.60) group than in the Non-AIBCES/Non-BCES group. The difference in surgery-related mortality between BCESs and Non-BCESs was smaller than between AIBCESs and Non-AIBCESs in the crude analysis. Fourth, the certification of both the institute and surgeon is strong factors affecting the incidence of anastomotic leakage, but the incidence of postoperative complications (pneumonia and recurrent laryngeal nerve palsy) did not differ among the four groups in the hierarchical multivariable logistic regression analysis.

Despite recent advances in surgical techniques and perioperative management, esophagectomy remains a highly invasive surgery associated with a high mortality rate and potentially fatal postoperative complications [1]. Several studies have demonstrated a strong association between operative volume (hospital volume) and surgery-related mortality [5‐9]. Nishigori et al. reported that the unadjusted operative mortality rate in hospitals performing 30 or more procedures per year was more than 3 times lower than in hospitals performing fewer than ten procedures annually [6]. Yoshida et al. reported that the most recent data in the NCD showed that a lower annual hospital volume of esophagectomies was a significant independent risk factor for surgery-related mortality [4]. The OR for hospitals with an esophagectomy volume of 18.4–43 per year was 0.43 compared to hospitals with a volume of 0.2–6.2 per year, and was 0.65 for hospitals performing 6.4–18.2 esophagectomies per year. These reports indicate that the centralization of esophageal surgery in Japan, such that hospitals perform more than 20–30 esophagectomies per year is most appropriate, as judged by the present social and medical situations of our country. One provision to be AIBCES is more than 50 surgeries for esophageal disease within a period of 5 years. This means that the minimum number of esophagectomies for esophageal cancers at an AIBCES is less than 10 per year, which is fewer than reported in the above reports. However, the actual number of esophagectomies per year at AIBCESs is larger. In this study, the average number of surgeries per year at AIBCESs was 27.8, as opposed to 3.7 at Non-AIBCES.

Why was the post‐esophagectomy mortality rate lower at AIBCESs than at Non‐AIBCESs? Ghaferi et al. reported that surgery-related mortality reflects the ability to avoid mortality from severe complications; that is, high–volume esophagectomy centers are less likely to exhibit “failure to rescue.” [10]. Other aspects of care associated with improved outcomes include the abilities of the nursing and staffing care teams [11]. In the present study, the incidence of anastomotic leakage was lower at AIBCESs, but that of recurrent laryngeal nerve palsy was higher. The incidence of postoperative pneumonia, which can lead to surgery-related mortality, was equal between AIBCESs and Non-AIBCESs. This demonstrates that the incidence of postoperative complications is not directly associated with surgery-related mortality. Recently, there have been reports, suggesting that the disparity in outcomes after complex surgery may not be due to differences in hospital volume, but to differences in the patients’ characteristics [12, 13]. Indeed, in our data, many of the patient’s clinical characteristics were significantly worse (high age, weight loss ≥ 10%, ASAPS ≥ 3, DM with insulin use, hypertension, serum albumin < 2.5 g/dl, and serum creatinine > 1.2 mg/dl) in patients treated at Non-AIBCESs. It was evident that a study with adjustments for risk and cancer stage was needed before drawing any conclusions. In the present study, therefore, we performed risk‐adjusted analyses, which led to the same conclusion.

The Japanese Society of Hepato‐Biliary‐Pancreatic Surgery (JSHBPS) established a board certification system for expert surgeons in the hepato‐biliary‐pancreatic (HBP) field and began certifying expert surgeons in 2008. Only two societies in Japan, the JES and JSHBPS, have established a board certification system for specialists with a high level of expertise in their field, which is often referred to as the “third floor.” For esophageal and HBP surgeons, it is first possible to be certified as a “Board Certified Surgeon” by the Japan Surgical Society (1st floor), then as a “Board Certified Surgeon in Gastroenterology” by the JSGS (2nd floor). A BECS is a specialist referred to as the third floor. In 2016, Miura et al. reported that the operative mortality rates after hepatectomies performed with participation of certified instructors or expert surgeons were significantly better than when this operation was performed without them (3.5% vs. 4.3%) [14]. Likewise, operative mortality rates were lower after pancreaticoduodenectomies performed with participation of certified instructors or expert surgeons (2.2% vs. 3.8%) [15]. This result is consistent with the earlier findings of McKay et al. [16], who reported that the 90‐day in‐hospital mortality rate after hepatic resection was 4.6% when performed by surgeons with HBP fellowship training and with a full year of training in HBP surgery, but was 6.3% when performed by surgeons with a surgical oncology fellowship, 7.2% when performed by surgeons with other subspecialty training, and 15.3% when performed by general surgeons.

In our data, although there was not a significant difference in the incidences of postoperative complications between esophagectomy by a BCES and Non-BCES, the surgery-related mortality rates did significantly differ (1.7% vs. 2.2%). However, the backgrounds of the patients treated were also different. Surprisingly, patients treated by BCESs were in better condition than those treated by Non-BCESs. This may be because a large fraction of the operations performed by BCESs are at AIBCESs, where patient selection is stricter than at Non-AIBCESs. This made risk‐adjusted hierarchical multivariable logistic regression analysis necessary. This analysis revealed that whether the surgeon was or was not a BCES did not directly influence surgery-related mortality. There is thus a difference between our findings and those previously reported by Miura et al. for HBP. Contributing to this difference is first the different affected organs (esophagus vs HBP). Second, Miura divided the patients based on participation of a surgeon certified in the HBP field, whereas we divided them based on whether the surgeon performing the operation was a BCES or Non-BCES. Consequently, the size of the BCES group was smaller than the Non-BCES group (6854 vs 9898). However, it is likely that even in the Non-BCES group, there was frequent participation of a BCES, particularly at AIBCESs [17]. In esophageal cancer surgery, the operation consists of resection in three fields (neck, thorax and abdomen) and reconstruction from the neck to the abdomen. The operation in the neck and abdomen is usually performed simultaneously in a high-volume hospital. In fact, there are multiple operating surgeons at work during esophageal cancer surgery. This is not registered directly in the NCD system, which permits only one operating surgeon to be listed. Third, we performed hierarchical multivariable logistic regression analysis that included not only the patients’ backgrounds, but also factors related to the institutions and surgeons, which Miura et al. did not. We think that it is for these reasons that factors related to the operating surgeon did not strictly influence postoperative complications.

Although it is clear from earlier studies that centralization of esophagectomy to high-volume hospitals improves short-term surgical results in patients with esophageal cancer, such centralization can cause problems, especially in local regional areas away from the cities. Surgery at a hospital far from the area where the patient lives is not suitable because of the characteristics of esophageal cancer. Esophageal cancer surgery is highly stressful for patients and requires intensive postoperative management. Moreover, esophagectomy requires pre- and postoperative adjuvant treatments. Consequently, the period of treatment is long. To achieve a complete cure, these patients require family and/or social support during recovery from the initial treatment and during the long-term follow-up. It is, therefore, necessary to consider how to balance centralization and equalization so as to optimize the medical delivery system not only for metropolitan areas but also for local regional areas away from the city. This research shows that the AIBCES certification system provided by the JES is fully functioning and maintained at a high level for esophageal cancer surgical treatment. There are differences in the best medical treatment flow for different diseases in different countries. There, too, medical societies for each specialized field establish appropriate medical delivery systems, and facility certification systems appear to play an important role, just as with the JES.

Despite the highly reliable data acquired via a robust statistical analysis of a very large cohort, this study has several limitations. First, there were differences among the esophagectomies registered in the NCD; that is, there were differences in the extent and intensity of lymph node dissection as well as in the route and organs used for reconstruction. Indeed, the incidence of recurrent laryngeal nerve palsy was higher in AIBCESs than Non-AIBCES (13.1% vs 11.4%). This may indicate a difference in the intensity of the bilateral cervical and upper mediastinal lymph node dissection between the two groups. Second, preoperative adjuvant therapy was not included in the risk factor adjustment. Using the NCD, Yoshida et al. showed that the surgery-related mortality was significantly higher in patients who received preoperative chemoradiotherapy than in those receiving only chemotherapy or no preoperative treatment [4]. In our earlier study, the fraction of patients receiving preoperative chemoradiotherapy were much smaller in Non-AIBCESs. Furthermore, Tsukada et al. reported that patients at lower volume hospitals were less likely to receive neoadjuvant therapy for esophageal cancer [18]. This tendency may influence the result of this study. Third, there is the possibility that unmeasured confounders influenced the results of these analyses.

In conclusion, our findings indicate that the institute certification system established by the JES is appropriate, as indicated by improved surgery-related mortality. Together with the surgeon and institute certification system, the JES is contributing to a more appropriate medical delivery system for thoracic esophageal cancer in Japan.