Japanese Journal of Clinical Oncology Advance Access originally published online on December 5, 2008
Japanese Journal of Clinical Oncology 2009 39(2):127-131; doi:10.1093/jjco/hyn138
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© The Author (2008). Published by Oxford University Press. All rights reserved
Trends of Centralization of Childhood Cancer Treatment Between 1975 and 2002 in Osaka, Japan
1 Department of Mathematical Health Science, Course of Health Science, Graduate School of Medicine, Osaka University
2 Department of Pediatric Hematology/Oncology Children’s Medical Center, Osaka City General Hospital
3 Department of Cancer Control and Statistics, Osaka Medical Center for Cancer and Cardiovascular Diseases, Osaka, Japan
For reprints and all correspondence: Yuko Ohno, Department of Mathematical Health Science, Course of Health Science, Graduate School of Medicine, Osaka University, 1-7 Yamadaoka, Suita, Osaka, Japan. E-mail: ohno{at}sahs.med.osaka-u.ac.jp
Received June 25, 2008; accepted November 1, 2008
 |
Abstract |
|---|
 TOP Abstract INTRODUCTIONPATIENTS AND METHODRESULTSDISCUSSIONFundingReferences |
|---|
Objective: To analyze the tendency to centralize childhood cancer treatment among cancer treatment hospitals in Osaka, Japan over a 28-year period.
Methods: The subjects were patients under the age of 15, newly diagnosed with cancer in Osaka between 1975 and 2002 (n = 4738). They were categorized into three groups by the time diagnosed (1975–84, 1985–93 and 1994–2002). The International Classification of Childhood Cancer was used as the disease classification. The degree of centralization was examined using a Pareto analysis, the Gini coefficient and the annual average number of cases per hospital.
Results: During this period, the number of children with cancer in Osaka has decreased by nearly half, from 2.1 to 1.2 million and the number of hospitals treating childhood cancer decreased from 37 to 20. However, the Pareto curve and Gini coefficient were almost constant (0.747, 0.737, 0.756 in Gini coefficient for the three diagnosed periods). The annual average numbers of cases per hospital were much low and marginally increased from 5.6 during 1975–84 to 6.1 during 1994–2002 in the hospitals that treated 90% of all cancers.
Conclusions: The degree of centralization appeared to be almost constant from 1975 to 2002 regardless of the decrease in hospitals treating cancer patients.
Key Words: childhood cancer • centralization • population-based cancer registry
|
INTRODUCTION |
|---|
|
TOPAbstractINTRODUCTIONPATIENTS AND METHODRESULTSDISCUSSIONFundingReferences |
|---|
Recent studies have suggested that there is a relation between better survival and hospital procedure volume (1–3). There also seems to be a relation between childhood cancer patients under the age of 15 referred to specialist centers and better survival (4–5).
The incidence rates of childhood cancer are very low and almost the same worldwide, and were 154 per million for boys and 158 per million for girls in Osaka in 2004 (6). So, centralization is thought to be particularly important to attain better survival for rare childhood cancers (4–5).
In fact, childhood cancer treatment has been centralized in the UK, Georgia and Germany (7–10), although the child and family are heavily burdened with ambulant treatment and hospital stays. For example, at 244 820 km2 and with 11.1 million children in 2001, the UK has only 22 treatment centers, which have treated 
90% of newly diagnosed cases (1350 cases per year in the early 2000s) (7–8). In Osaka with an area of 1898 km2 and 1.3 million children in 1995, nine specific hospitals had treated 70% of cases (180 cases per year) during 1989–98 (11). This information suggests that childhood cancer treatment had been decentralized in Osaka (7–10).
The child population in Osaka has reduced by about half, from 2.1 million in 1975 to 1.2 million in 2005, because of a decreasing birth rate. According to this decline, the number of childhood cancer patients and the annual treatment volume per hospital should also have decreased. On the other hand, childhood cancer treatment was supposed to have been centralized to specific hospitals in 1990s mainly because of following three factors. First, the number of medical lawsuits increased and this led doctors to refer patients that were difficult to diagnose to specialists (12). Second, multi-center cooperative studies began to conduct larger scale investigations into childhood cancer so that patients could be treated at specific hospitals (13). Third, two large and specialized hospitals for pediatric medicine were built and a university hospital relocated to Osaka in the early 1990s.
We investigated the trend of centralization of cancer patients to specific hospitals from 1975 to 2002 using the data from the population-based Osaka Cancer Registry (OCR).
|
PATIENTS AND METHOD |
|---|
|
TOPAbstractINTRODUCTIONPATIENTS AND METHODRESULTSDISCUSSIONFundingReferences |
|---|
Data Sources
Osaka Prefecture is an area with 8.8 million people, and the population covered by the OCR was the largest in Japan till 2006. The proportion of death certificate only (DCO) cases was <2.9% for children under the age of 15 between 1981 and 89, and that of registered patients was estimated to be from 85 to 94% (14). We used this OCR as our source database.
Subjects
Our subjects were patients under the age of 15 and newly diagnosed with cancer between 1975 and 2002 (5291 cases). We then excluded carcinoma in situ (two cases), patients who had no information on a treatment (153 cases), those who did not specify both treating hospital code and diagnosing hospital code (three cases), and those that received treatment in other prefectures (417 cases). The final number of subjects for our study was 4738.
We needed to know the number of patients by treatment hospital in order to investigate the degree of centralization. For the analysis, we principally used the treating hospital code (3741 cases). The diagnosing hospital codes were used only when the patients had no information on the treating hospital (997 cases).
The patients were then divided into three period groups by the diagnosed year: 1975–84 (1976 cases), 1985–93 (1661 cases) and 1994–2002 (1101 cases). We then used these data to investigate the centralization tendency.
Classification of Childhood Cancer
Childhood cancer is histologically very diverse and some histological types occur in many different sites. Since 1996, the International Classification of Childhood Cancer (ICCC) has been used for international comparisons of statistical analyses (15). We used twelve diagnostic groups on the basis of ICCC in this research for our classification of childhood cancer.
Method
A Pareto analysis was used to investigate the centralization tendency for childhood cancer treatment. The process for the Pareto analysis was as follows.
A table was prepared to list the number of patients that each hospital treated, and the rows were arranged in descending order of the number of the patients. A column was added to this table that shows the cumulative percentage of the patients in descending order.
In order to find how many hospitals have treated childhood cancers, we counted up the cumulative frequency of hospitals when the cumulative percentage of patients was upper 50 and 75% and when it was 100% by year, diagnosed period and ICCC. We also plotted the Pareto curves with the cumulative percentage of hospitals and patients for the three diagnosed periods.
The Gini coefficient was also calculated as an index of the centralization of treatment for the three diagnosed period groups and the ICCC group. This is defined as follows (16):
|
|
Graphically, the Gini coefficient covers twice the area between the Pareto curve and the line of equality. The Gini coefficient ranges between 0 and 1. A ‘one’ means all the patients were treated at one specific hospital.
Furthermore, the annual average number of cases per hospital was calculated and compared with those of European countries and the USA (8–10).
We used R version 2.4.0 for all the analysis in this study.
|
RESULTS |
|---|
|
TOPAbstractINTRODUCTIONPATIENTS AND METHODRESULTSDISCUSSIONFundingReferences |
|---|
The number of children in Osaka during this 28-year period decreased to nearly half, from 2.1 to 1.2 million. That of the hospitals treating childhood cancer also decreased from 37 in 1975 to 20 in 2002, and in particular, decreased from 37 in 1993 to 32 in 1994. After 1993, the number of the newly diagnosed patients drastically decreased from 161 in 1993 to 107 in 2002.
Figure 1 shows the change in the Pareto curve and the Gini coefficient for all childhood cancer cases for the diagnosed period. The Gini coefficients were almost constant (0.747, 0.737, 0.756 for the three periods), and a weak centralization can be found during 1994–2002 when looking at the Pareto curve.
|
Table 1 shows each cumulative frequency of hospitals and the Gini coefficient by the ICCC and diagnosed period. Each cumulative frequency of hospitals slightly increased or was constant during 1975–84 and 1985–93, but reduced by about half during 1994–2002, except for the hospitals treating upper 50% of malignant bone tumors (VIII). The variations in Gini coefficient were small throughout all three periods. The Gini coefficients by the ICCC groups ranged from 0.275 to 0.665. The rarer the ICCC group was, the smaller the Gini coefficient was. In particular, the low level of centralization was shown for the rarely diagnosed groups: retinoblastoma (V), renal tumors (VI), hepatic tumors (VII), malignant bone tumors (VIII), carcinomas and other malignant epithelial neoplasms (XI) and other and unspecified malignant neoplasms (XII).
|
The annual average numbers of cases per hospital were less 2.2 in all hospitals during the three diagnosed periods (1975–84: 2.17, 1985–93: 2.15, 1994–2002: 2.18). In hospitals that treated upper 90% of all cancers, those were 5.6 during 1975–84, 5.5 during 1985–93 and 6.1 during 1994–2002. In the early 2000s, those were
61.4 in the UK, 60.9 in Georgia and 32.4 in Germany (8–10). Even hospitals that treated upper 50% of all cancers, the annual average numbers of cases were 14.1 during 1975–84, 15.4 during 1985–93 and 20.4 during 1994–2002. Those were one-third the number compared with 56.3 in Germany (10). In addition, during 1994–2002, for 10 out of 12 ICCC groups, the annual average number of cases per hospital was less than one person. |
DISCUSSION |
|---|
|
TOPAbstractINTRODUCTIONPATIENTS AND METHODRESULTSDISCUSSIONFundingReferences |
|---|
The trend of centralization of childhood cancer treatment was studied based on the population-based cancer registry in Osaka. Little was previously known about this trend. Our study is valuable because we used population-based data that had about an 85–94% registration rate for children, and we looked at the information for a 28-year period (14).
After 1993, each cumulative frequency of hospitals decreased to about half along with the number of newly diagnosed cancers, and the degree of centralization was almost constant during the three diagnosed periods. These results suggest that childhood cancers have continually been treated at many different hospitals, and the reduction in the number of cases also reduced each cumulative frequency of hospitals. From 1993 to 1994, each cumulative frequency of hospitals reduced approximately by one-third. Around 1993, three large hospitals completed or relocated, so this was supposed to have influenced the trend of hospitals that accessed childhood cancer.
The annual number of cases by the ICCC deserves attention in terms of absolute smallness (Table 1). Despite this smallness, patients were treated at many different hospitals. And the annual average number of cases per hospital has remained very small, which was much lower compared to that in the UK, Georgia and Germany. In addition, each diagnostic group includes many types of cancer and their different treatments. So, centralization to specific hospitals is necessary to improve survival.
Looking at the results, the rarer the ICCC group was, the smaller the Gini coefficient was. Of the types of cancer found in children, clinical trials have been mainly conducted on the more common tumors, such as leukemia (I) or lymphomas and reticuloendothelial neoplasms (II) (13,17). This would have helped to more centralize the more common types of childhood cancer. More influential reasons for doing this would be as follows. Patients were able to freely select and attend treatment hospitals in the Japanese medical system, although they did not usually have enough information on where were specialists. Of total 566 hospitals in Osaka, patients had to find an appropriate one under uncertainty and incomplete information (18). Regretfully, general practitioners had not strictly referred cancer patients to childhood cancer specialists.
From a statistical point of view, the Gini coefficient for a small sample is known to include a downward bias (19,20). The Gini coefficients of each category may have a downward bias, and look to lower centralization, particularly during 1994–2002. Each cumulative frequency of hospitals for sympathetic nervous system tumors (IV) was exceptional compared with those of the other diagnostic groups. This was due to the mass screening at 6 months of age for neuroblastoma (IVa) that was introduced in 1985 and had continued into the 2000s across Japan. As a result of over-diagnosis, the annual age-standardized incidence rate increased by about three times in children from the periods of 1970–84 to 1985–94 (21). Table 1 also shows the increase in the total number of cases of sympathetic nervous system tumors (IV). However, as <10 hospitals carried out mass screening in Osaka, each cumulative frequency of hospitals during 1985–93 is about equal to those during 1994–2002.
With the exception of neuroblastoma (IVa), the survival of many diagnostic groups in Osaka was lower than that in England and Wales and in the USA, and the report suggested that the reason for this was insufficient introduction and practice of chemotherapy (22). Our study suggested that the low centralization of patients is also related to the lower survival. Previous studies also suggested that the lower survival will also related to the treatment volume in the field of surgery or radiotherapy, although these subjects were adult cancers (23,24). Therefore, to centralize childhood cancer to specific hospitals and to perform a higher volume of procedures are important to ensure better survival. For the centralization of treatment, however, the burden children and their families must deal within their daily lives would increase. A social support system would be needed to achieve and maintain centralization.
In our study, although the identification of treating hospitals was the point, 997 cases did not have treating hospital codes so that the diagnosing hospital code was alternatively adopted. The bias derived from this substitution is assumed minor, because the proportion of cases that the diagnosing hospital code was same as the treating hospital one was 91.8% (3435 cases).
These data included newly diagnosed patients only, so that the specialists might feel that the small degree of centralization would not reflect the realization for childhood cancer treatment. Further study would be needed to investigate the centralization taking into account the succession of treatment.
We confirmed that the hospitals that treated childhood cancers decreased approximately by half during the 1990s, because childhood cancer decreased because of a lower birth rate. The degree of centralization seemed almost constant from 1975 to 2002. The annual average number of cases per hospital marginally increased, although it still was much lower compared with European countries and the USA.
|
Funding |
|---|
|
TOPAbstractINTRODUCTIONPATIENTS AND METHODRESULTSDISCUSSIONFundingReferences |
|---|
This study was supported by a Health and Labour Sciences Research Grant for Clinical Cancer Research (H19-009) and a Grants-in-Aid for Scientific Research (B) (18310105) and a Cooperative Research from the Institute of Statistical Mathematics.
Conflict of interest statement
None declared.
|
References |
|---|
|
TOPAbstractINTRODUCTIONPATIENTS AND METHODRESULTSDISCUSSIONFundingReferences |
|---|
1 Dickstein RJ, Barone JG, Liao JG, Burd RS. The effect of surgeon volume and hospital characteristics on in-hospital outcome after ureteral reimplantation in children. Pediatr Surg Int (2006) 22:417–21.[CrossRef][Web of Science][Medline]
2 Killeen SD, O’sullivan MJ, Coffey JC, Kirwan WO, Redmond HP. Provider volume and outcomes for oncological procedures. Br J Surg (2005) 92:389–402.[CrossRef][Web of Science][Medline]
3 Ioka A, Tsukuma H, Ajiki W, Oshima A. Hospital procedure volume and survival of cancer patients in Osaka, Japan: a population-based study with latest cases. Jpn J Clin Oncol (2007) 37:544–53.
4 Stiller CA. Centralisation of treatment and survival rates for cancer. Arch Dis Child (1988) 63:23–30.
5 Stiller CA, Draper GJ. Treatment centre size, entry to trials, and survival in acute lymphoblastic leukaemia. Arch Dis Child (1989) 64:657–61.
6 Osaka Prefectural Department of Public Health and Welfare, Osaka Medical Association, Osaka Medical Centre for Cancer and Cardiovascular Diseases. Annual Report of Osaka Cancer Registry No. 71 - Cancer Incidence and Medical Care in Osaka in 2004 and the Survival in 2000: OPDPHW. (2008).
7 Children’s Cancer and Leukaemia Group [homepage on the Internet]. Leicester: Children’s cancer, Inc.; c2006-08 [cited 2008 Aug 27]. Available from: http://www.ukccsg.org/.
8 Ablett S, Pinkerton CR. Recruiting children into cancer trials–role of the United Kingdom Children’s Cancer Study Group (UKCCSG). Br J Cancer (2003) 88:1661–5.[CrossRef][Web of Science][Medline]
9 Howell DL, Ward KC, Austin HD, Young JL, Woods WG. Access to pediatric cancer care by age, race, and diagnosis, and outcomes of cancer treatment in pediatric and adolescent patients in the state of Georgia. J Clin Oncol (2007) 25:4610–5.
10 Kaatsch P, Spix C. German Childhood Cancer Registry, Annual Report 2003. (2004) Mainz: German Childhood Cancer Registry.
11 Ajiki W, Tsukuma H, Oshima A. Step toward the realization of a population-based national childhood cancer registry in Japan: recommendation from population-based cancer registry. Jpn J Pediatr Oncol. (2002) 39:150–8. (in Japanese).
12 Kobayashi H. Risk-management in Pediatric wards. Juntendo Med J (2005) 51:490–500. (in Japanese).
13 Makimoto A, Kawamoto H. Future strategy for developing childhood cancer treatment (Featuring article Recent outcomes and problems in childhood cancer). Pediatr Jpn (2005) 46:1995–2002. (in Japanese).
14 Ajiki W, Hanai A, Tsukuma H, Fujimoto I, Oshima A, Parkin DM. International Incidence of childhood cancer, vol.II (IACR scientific publications; 144) (1998) Lyon: the International Agency for Research on Cancer. 189–91.
15 Kramárová E, Stiller CA. The international classification of childhood cancer. Int J Cancer (1996) 68:759–65.[CrossRef][Web of Science][Medline]
16 Gini C. Sulla misura della concentrazione e della variabilità dei caratteri. Atti del Reale Istituto Veneto di Scienze, Lettere ed Arti, A.A. 1913–1914. (1914) 73:1203–48.
17 Makimoto A. Infrastructure development for establishing the clinical traials for child solid tumors in Japan (Featuring article New treatment of solid tumors). J Pediatr Pract (2004) 67:577–82. (in Japanese).
18 Statistics and Information Department, Minister’s Secretariat, Ministry of Health, Labor and Welfare. (2002) Survey of Medical Care Institutions. Health, Labor and Welfare Statistics Association. Tokyo (Japan): Health, Labor and Welfare Statistics Association; 2004 (in Japanese).
19 Lerman RI, Yitzhaki S. Improving the accuracy of estimates of Gini coefficients. J Econ (1989) 42:43–7.
20 George D. The small-sample bias of the Gini coefficient: results and implications for empirical research. Rev Econ Stat (2003) 85:226–34.[CrossRef][Web of Science]
21 Ajiki W, Tsukuma H, Oshima A, Kawa K. Effects of mass screening for neuroblastoma on incidence, mortality, and survival rates in Osaka, Japan. Cancer Causes Control (1998) 9:631–6.[CrossRef][Web of Science][Medline]
22 Ajiki W, Tsukuma H, Oshima A. Survival rates of childhood cancer patients in Osaka, Japan. Jpn J Clin Oncol (2004) 34:50–4.
23 Tanisada K, Teshima T, Ohno Y, Inoue T, Abe M, Ikeda H, et al. Patterns of Care Study quantitative evaluation of the quality of radiotherapy in Japan. Cancer (2002) 95:164–71.[CrossRef][Web of Science][Medline]
24 Saika K, Ohno Y, Tanaka H, Hasegawa T, Tsukuma H, Oshima A. The trend of the effect of surgical volume for stomach and lung cancer patients. Jpn J Appl IT Healthc (2007) 1:42–9.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||