1 INTRODUCTION

Cancer is associated with the growing burden of disease in public health. Globally, 18.1 million people had cancer, and 9.6 million died from the disease in 2018.¹ Cancer control to reduce the burden of cancer mainly comprises primary prevention, screening, early diagnosis, treatment, survivorship, and palliative care.¹ Of these interventions, treatment is largely responsible for patient outcomes. The centralization of patients at medical facilities equipped with better infrastructure, resources, knowledge, and experience is a common strategy to improve the quality of care, patient outcomes, and financial efficiency.^{2, 3} The concept of centralization stems from the evidence that higher-volume hospitals are associated with better patient outcomes.⁴ For example, in cancer treatment, patients treated in high-volume hospitals showed higher 5-year survival rates^{5, 6} and lower mortality rates⁷ than those treated in low-volume hospitals. Although the underlying mechanism of this association has not been established,⁸ high-volume hospitals may be more capable of practicing specific surgical procedures and intensive care and managing postoperative complications than low-volume hospitals.^{9, 10}

Several countries define the minimum surgical volume standards as a part of the centralization strategy.^{2, 11} For example, of 22 European and North American countries, more than half have surgical volume standards for esophageal, pancreatic, liver, or rectal cancers, but the cutoffs of volume standards vary with cancer site and country.² Some studies have reported that hospitals that meet surgical volume standards report lower mortality than those that do not meet the standards.^{12, 13} In contrast, some studies found no association, calling for the need to identify the appropriate requirements.^{14, 15} This implies that the surgical volume standard should be defined according to the health system, regulation, cancer burden,¹¹ or training system of the surgical experts in local settings, and some studies have investigated the minimum surgical volumes for specific cancer sites based on the associations with patient outcomes.^16-18

In Japan, cancer has been the leading cause of death since 1981, accounting for 27% of all-cause deaths in 2018.¹⁹ The national government addresses nationwide cancer control, namely the Basic Plan to Promote Cancer Control programs,²⁰ and developed cancer care system which placed Designated Cancer Care Hospitals (DCCHs) at the center of cancer treatment and cares to ensure patients with cancer access high-quality cancer care regardless of where they live.²¹ As of 2021, 405 DCCHs were designated by the national government (i.e., the national DCCHs) across the country.²² In addition, prefectural governments designate prefectural DCCHs to strengthen the local cancer care system.²³ Both national and prefectural DCCHs need to meet various standards for structure, resource, and clinical practice, including annual surgical volume for all cancers.^{24, 25} However, surgical volume standards per cancer site are not defined for DCCHs. Meanwhile, medical societies or associations define surgical volume standards to certify hospitals as high-standard medical or training facilities. For example, the required surgical volumes are ≥50 esophageal surgeries within 5 years,²⁶ ≥60 surgical gastrectomies for gastric cancer within the last 3 years,²⁷ ≥100 colorectal and anal surgeries per year,²⁸ average ≥ 20 pancreatic cancer cases per year registered in the national clinical database within the last 5 years,²⁹ average ≥ 150 thoracic surgeries per year in the last 3 years to certify flagship facilities for specialized training, an average of ≥25 thoracic surgeries per year in the last 3 year to accredit collaboration facility for specialized training,³⁰ and ≥30 surgeries for breast cancers annually.³¹

The association between high surgical volume and better patient outcomes is known in Japan,^32-35 and minimum surgical volume thresholds for solid cancers based on the association with patient survival probability have been previously reported.³⁶ However, the minimum surgical volume for specific cancer sites has not been examined based on the volume–outcome relationship. Therefore, this study aimed to identify the minimum surgical volume per cancer site based on the association between surgical volume and adjusted 5-year survival probability.

2 MATERIALS AND METHODS

3 RESULTS

3.4 Minimum surgical volume based on the associations with the adjusted 5-year survival probability

Figure 1 shows the trends of adjusted 5-year survival probability per increase of five surgical volumes and their joinpoints, which we considered the suggested minimum surgical volume. The suggested surgical volumes were identified as 35–39, 20–24, 25–29, 10–14, 10–14, and 25–29 for esophageal, stomach, colorectal, pancreatic, lung, and breast cancers, respectively. The slope below and above the suggested surgical volume (by percent points), which indicates the extent of change in the adjusted 5-year survival probability per increase of five surgeries, was +2.23 and + 0.39 for the esophagus, +9.68 and + 0.34 for the stomach, +8.11 and + 0.05 for the colorectum, +3.82 and + 0.87 for the pancreas, +9.46 and + 0.23 for the lung, and + 1.27 and + 0.03 for the breast.

Figure S1 shows the trends of adjusted 5-year survival probability per increase of five surgical volumes and their joinpoints which excluded patients with metastatic or unknown cancer stages (supporting information). The suggested surgical volumes were identified as 35–39, 20–24, 25–29, 10–14, 30–34, and 20–24 for esophageal, stomach, colorectal, pancreatic, lung, and breast cancers, respectively.

4 DISCUSSION

This study examined the association between surgical volume and adjusted 5-year survival probability for six selected cancers and estimated the suggested minimum surgical volume based on the association with survival probability among DCCHs in Osaka. Although DCCHs are specialized in cancer treatment and care, DCCHs with higher surgical volume appeared to show a higher adjusted 5-year survival probability than those with lower volume, as several studies have reported the volume-outcome relationship. To the best of our knowledge, this is one of the few studies^16-18 that have identified the suggested minimum surgical volume per cancer site. The surgical volumes of the DCCHs showed a wide range per cancer site, with maximum surgical volumes of 55–59, 150–154, 195–199, 30–34, 160–164, and 195–199 for esophageal, stomach, colorectal, pancreatic, lung, and breast cancers, respectively. Within these ranges, the suggested surgical volumes were identified at 35–39, 20–24, 25–29, 10–14, 10–14, and 25–29 surgeries per year for esophageal, stomach, colorectal, pancreatic, lung, and breast cancers, respectively, after adjustment for patient characteristics. Upward trends in the adjusted 5-year survival probability per increase in surgical volume were observed until the suggested surgical volume thresholds across the six cancer sites. After that, the trends nearly plateaued, as similar trends were observed in previous studies.^{46, 47} However, the slopes for the 5-year survival probability per increase in surgical volume varied with cancer site. Further, there could be no specific minimum surgical volume identified if the slopes before and after the joinpoint were flat or constant (i.e., liner trend). Thus, various aspects should be considered for the application of the minimum surgical volume in policy and clinical settings, such as the slopes before and after the suggested minimum surgical volume, hazard ratios, and opinions of surgical specialists.

Of the six selected cancers assessed, the stomach, colorectal, lung, and breast cancers were the major cancers in Osaka. Our results showed that the slopes of the associations between surgical volume and the adjusted 5-year survival probability for stomach, colorectal, and lung cancers showed steep trends until the suggested surgical volume, which flattened subsequently. This indicates that patients treated at lower-volume DCCHs show lower survival probability, which suggests that monitoring and strengthening low-volume DCCH function will be important to improve patient outcomes. This is because the DCCH system was developed to offer high-quality cancer medicine to patients with these common cancers within their residential areas.²¹ To improve the performance of low-volume DCCHs, availability of physician specialty training and intensive care units, availability of certain clinical services, staff-patient ratio, and adherence to evidence-based care should be considered, as these are prioritized in high-volume hospitals.⁸ Breast cancers showed a minimum disparity in the adjusted 5-year survival probability by surgical volume, and the slopes before and after the suggested surgical volume threshold were not steep, probably because the survival probability of breast cancer is higher than that of other cancers. Moreover, the effect of surgery on the survival probability of breast cancers may be relatively smaller than that of other cancers because chemotherapy and radiation therapy also play critical roles in treating breast cancers.

More than 90% and 80% of DCCHs that performed surgeries for esophageal and pancreatic cancers, respectively, reported less than the suggested surgical volume; 54% and 46% of patients with esophageal and pancreatic cancers were surgically treated at DCCHs with volumes below the suggested surgical volume, although the estimated joinpoint regression was less likely to be robust than that for other cancer sites because of the smaller sample, which requires careful interpretation. Despite this, our result shows that patients with these cancers are scattered across low-volume DCCHs, although the number of patients was lower than that for other cancers. It is likely that many patients with esophageal and pancreatic cancers may first access a hospital near their places of residence, which may not necessarily be a DCCH with sufficient surgical volume. However, these hospitals may not necessarily have sufficient experience with esophageal and pancreatic surgeries. Furthermore, most patients were diagnosed with regional- or distant-staged cancers, which were associated with poor prognoses. However, physicians or patients may not choose referrals to higher-volume DCCHs. Similarly, in the United States and Germany, a high proportion of patients treated in hospitals that do not meet surgical volume standards have been reported.^{12, 14, 48} These findings suggest that centralizing patients with esophageal and pancreatic cancers at higher-volume DCCHs and developing better surgical or non-surgical treatment options will improve their survival probability, as the centralization policy has successfully improved patient outcomes with pancreatic cancer in the Netherlands⁴⁹ and those with esophagogastric cancer in England.⁵⁰ Meanwhile, centralization may increase travel distance to hospitals⁵¹ and create barriers for vulnerable patients to receive care in high-volume hospitals.⁵² In Osaka, these would be less critical concerns because it has a well-developed transportation system and 66 DCCHs in operation despite the small-sized prefecture, and nearly, the entire population in the country is insured.⁵³

We compared our suggested surgical volume with existing surgical volume standards by the Japanese medical societies or associations (Table 2). Regarding Japanese medical societies and associations, surgical volume standards for stomach, pancreatic, and breast cancer surgeries are used for credentialing hospitals, but those specific for esophageal, colorectal, and lung cancers are not available. The suggested minimum surgical volume of this study and the existing surgical volume standards per year were as follows: 20–24 versus 20 for stomach cancer,²⁷ 10–14 versus 20 for pancreatic cancer,²⁹ and 25–29 versus 30 for breast cancer,³¹ respectively. This demonstrates that the existing surgical volume standards are almost equal to or slightly higher than the suggested surgical volume.

The suggested minimum surgical volume was not necessarily consistent with the surgical volume standards in other settings. To investigate this, we compared our suggested surgical volume with the surgical volume standards used in the US and Germany (Table 2). Our suggested surgical volume and the standards of the US Leapfrog organization were as follows: 35–39 versus 20 for esophagus cancer,⁴¹ 10–14 versus 20 for pancreatic cancer,⁴¹ and 10–14 versus 40 for lung cancer.⁴¹ We also compared our suggested surgical volume with the surgical volume standards for the certified cancer center program of the German Cancer Society; our suggested surgical volume and the standards were as follows: 25–29 versus 50 for colorectal cancer (e.g., 30 colon and 20 rectal cancer surgeries),⁵⁴ 10–14 versus 12 for pancreatic cancer,⁵⁵ and 10–14 versus 75 for lung cancers.⁵⁶ These comparisons showed that the extent of the similarity between our suggested surgical volume and the existing standards in the US and Germany varies by cancer site. These differences are not surprising because they imply that estimates of the minimum surgical volume are affected by incidence, prognosis, and other contextual factors in local settings; thus, a single surgical volume threshold is not necessarily suitable for all contexts.

This study had several limitations. First, the hazard ratios and survival probabilities may be biased because potential confounders that were unavailable in our cancer registry were not adjusted. They are socioeconomic characteristics, comorbidity, performance status, time from diagnosis to surgery, and any disease and treatment that a patient had after surgery for cancer. Other hospital characteristics (e.g., surgeon volume, human and material resources, or infrastructure) were not adjusted. Second, the cancer-specific mortality rate could not be estimated because of the absence of information on the cause of death in the registry. Finally, the generalizability of the study findings is limited because the burden of each cancer and existing regulations and health systems may vary in other parts of the country and the world. However, Osaka prefecture has a high-quality population-based cancer registry and a well-functioned designated cancer care hospital system. Thus, our study findings may reflect the situation of urban prefectures with well-functioning designated cancer care hospital system. Further, future research that estimates minimum surgical volume using nationwide data will be worthwhile when the data become officially available.

In conclusion, this study identified and suggested the minimum surgical volume of hospitals designated for cancer care. The identified surgical volume varied with the cancer site. Although a surgical volume standard for all cancers has been applied to assess the performance of DCCHs, cancer site-specific surgical volume standards may have clinical implications for improving the quality of surgery and reducing preventable complications and deaths. Further centralization of patients with esophageal and pancreatic cancers at high-volume DCCHs should be encouraged. In addition, strengthening the capability of low-volume DCCHs to treat common cancers is also important to ensure that all patients receive high-quality treatment within their residential area and, consequently, improve their survival.

AUTHOR CONTRIBUTIONS

SO and IM conceived and designed the study. SO conducted the statistical analysis. SO, TT, and MI interpreted the data. SO drafted the manuscript. All authors contributed to the critical revision of the manuscript and approved the final version of the manuscript. IM supervised the study.

CONFLICT OF INTEREST

Dr. Isao Miyashiro is a member of the Cancer Registry Data Use Review Committee, which evaluates the confidentiality of study subjects. His work as a committee member does not affect any part of this study.

ETHICS STATEMENT

Ethical approval was obtained from the Institutional Review Board of the Osaka International Cancer Institute (approval number: 19143, 19144) before conducting the study. We obtained the dataset with no personally identifiable information from the Osaka Cancer Registry and independently processed it according to the Act on the Promotion of Cancer Registries. Informed consent was not required for the patients.