Japanese Journal of Clinical Oncology

Simulation of National Average Using Structure Survey by Tsunemoto (6)

A calculation program was written by means of SAS 6.12 (7,8) according to the methodology mentioned above. National averages for the radiotherapeutic processes used for esophageal cancer patients who received radiotherapy in Japan between 1992 and 1994 were calculated. The items calculated were whether patients had undergone endoscopic ultrasound examination, whether all fields were irradiated each day and whether patients were being treated with brachytherapy. These were typical items that had apparent differences by the stratification of institution and were important for radiotherapeutic processes represented by binary data.

Patient data, y_hik, were obtained from the PCS for esophageal cancer patients in Japan carried out by our group from July 1996 to February 1997. External audits were performed by the principal investigators (M.A. and H.I.) of the cancer research groups on the member institutions who agreed to official requests for this survey by the Ministry of Health and Welfare in Japan (8-27 and 8-29) and their affiliated hospitals, as shown in Table 1. The number of institutions audited was 29 and the fifth PCS data format produced in the USA was used. The total number of esophageal cancer patients surveyed was 455. These included patients who received definitive, palliative, preoperative or postoperative radiotherapy. Patients who had distant metastasis or were diagnosed with any other cancer within 5 years before the start of treatment were excluded.

Japanese institutions of radiation oncology were stratified into four categories according to the facility master survey by Tsunemoto (6). On the basis of this stratification, 39 institutions were classified as A1 (university hospital/cancer center treating [ge]300 patients/year), 51 institutions as A2 (university hospital/cancer center treating <300 patients/year), 87 institutions as B1 (other institutions treating [ge]120 patients/year) and 197 institutions as B2 (other institutions treating <120 patients/year) in the current series. This stratification of institutions was closely related to the number of full-time equivalent radiation oncologist (4,9). The actual number of sampled institutions, n_h, was 13 for A1, 0 for A2, 10 for B1 and six for B2 (Table 2).

Table 1. Institutions audited and responsible persons and staff who collaborated in this study

Institution	Location	Responsible person and staff
Sakai Municipal Hospital	Sakai	Hiroyasu Yoshioka
Suita Municipal Hospital	Suita	Masayuki Sato
Sumitomo Hospital	Osaka	Jun Ueda
National Sapporo Hospital	Sapporo	Masamichi Nishio
Aomori Prefectural Central Hospital	Aomori	Sadao Watanabe Yoshinao Abe
Tohoku University Hospital	Sendai	Shogo Yamada Yoshihiro Takai
Niigata University Hospital	Niigata	Kunio Sakai Tadashi Sugita
National Kyoto Hospital	Kyoto	Mitsuyuki Abe Toru Shibata
Kyoto University Hospital	Kyoto	Masahiro Hiraoka Yasumasa Nishimura Yasushi Nagata
Kyushu University Hospital	Fukuoka	Kouji Masuda Satoru Uehara Junichi Omagari
Hiroshima University Hospital	Hiroshima	Yutaka Hirokawa Yukio Akagi
Hamamatsu University Hospital	Hamamatsu	Masao Kaneko Tetsuo Nishimura
Shinshu University Hospital	Matsumoto	Shusuke Sone Masahiko Oguchi
Gunma University Hospital	Maebashi	Hideo Niibe Norio Mitsuhashi Michitaka Yamakawa
National Cancer Center Hospital	Tokyo	Hiroshi Ikeda Yoshikazu Kagami Minako Sumi
Tokai University Hospital	Isehara	Tomoyuki Mori Yukio Oizumi
Osaka University Hospital	Suita	Toshihiko Inoue Takehiro Inoue
Cancer Institute Hospital	Tokyo	Takashi Yamashita Masahiko Furukawa Masao Kobayashi Hiroshi Igaki
National Osaka Hospital	Osaka	Masanori Mitomo Masatoshi Ohtani
Osaka Rosai Hospital	Sakai	Isao Tsukaguchi
Osaka Red Cross Hospital	Osaka	Giro Todo
Kansai Denryoku Hospital	Osaka	Daizaburou Hamanaka
Kyoto City Hospital	Kyoto	Katsumi Hayakawa Mototsugu Koishi
Hiroshima Red Cross and Atomic Bomb Survivors Hospital	Hiroshima	Masaki Mori Kazuki Kashimoto
Chuden Hospital	Hiroshima	Akira Naito
Onomichi General Hospital	Onomichi	Tetsuji Kiso
Chugoku Rosai Hospital	Kure	Katsuro Hanaguri
Nagano Red Cross Hospital	Nagano	Youichi Okazaki
Hokushin General Hospital	Nakano	Kiyonobu Ito

Table 2. Number of institutions at each stratum, median number of esophageal cancer patients by facility master survey in Japan by Tsunemoto in 1990 (6) and number of institutions audited from 1996 to 1997 (4,5)

Stratum	Total No. of patients per year	Nh	Mhi	nh
A1	Univ/CC >=300 patients	39	30	13
A2	Univ/CC <300 patients	51	13	0
B1	Others >=120 patients	87	7	10
B2	Others <120 patients	197	4	6

N_h, number of institutions at each stratum in 1990; M_hi, median number of esophageal cancer patients at each stratum in 1990; n_h, number of audited institution at each stratum; Univ, University Hospital; CC, Cancer Center Hospital. N_h and M_hi were derived from the structure survey by Tsunemoto in 1990 (6).

Table 3. Average and variance at each stratum and national average (All) of the process for esophageal cancer patients on the application of endoscopic ultrasound

Stratum	Total No. of patients per year	Average	Variance
A1	Univ/CC >=300 patients	0.280	0.0033
B1	Others >=120 patients	0.027	0.0004
B2	Others <120 patients	0.030	0.0004
All		0.129	0.0004

Univ, University Hospital; CC, Cancer Center Hospital.

Table 4. Average and variance at each stratum and national average (All) of the process for esophageal cancer patients on `all fields treated each day'

Stratum	Total No. of patients per year	Average	Variance
A1	Univ/CC >=300 patients	0.755	0.0080
B1	Others >=120 patients	0.561	0.0200
B2	Others <120 patients	0.412	0.0318
All		0.599	0.0058

Univ, University Hospital; CC, Cancer Center Hospital.

Table 5. Average and variance at each stratum and national average (All) of the process for esophageal cancer patients on the application of brachytherapy

Stratum	Total No. of patients per year	Average	Variance
A1	Univ/CC >=300 patients	0.166	0.0008
B1	Others >=120 patients	0.051	0.0004
B2	Others <120 patients	0.019	0.0003
All		0.088	0.0002

Univ, University Hospital; CC, Cancer Center Hospital.

Table 6. Simple average of the process for esophageal cancer patients by PCS between 1992 and 1994 in Japan (4)

	Simple average	Variance
Endoscopic US	0.158	0.1335
All fields treated each day	0.648	0.2286
Brachytherapy	0.117	0.1035

PCS, Patterns of Care Study; US, ultrasound.

Table 7. Number of institutions where it is possible to carry out brachytherapy in audited institutions

Stratum	Total No. of patients per year	No. of institutions with capability for brachytherapy
A1	Univ/CC >=300 patients	13 (100%)
B1	Others >=120 patients	7 (70%)
B2	Others <120 patients	0 (0%)

Univ, University Hospital; CC, Cancer Center Hospital.

The total number of institutions in each stratum, N_h, and the number of patients at each institution, M_hi, could not be identified with the PCS database format we used for the patients between 1992 and 1994. Therefore, data from the structure survey by Tsunemoto in 1990 (6) were used as a makeshift solution for the calculation of national averages. For the median number of patients we used the number of patients per year at each institution, M_hi (Table 2). Because data for esophageal cancer patients treated between 1992 and 1994 were used, the median number of patients in 1990 was multiplied by three.

RESULTS

The program for calculating national averages, written by means of SAS 6.12, proved to be successful and the pilot calculation of national averages for radiotherapy in Japan was simulated by using this program.

DISCUSSION

PCS is a widely known QA program for radiation oncology in the USA. The usefulness of the PCS is that it could retrospectively collect detailed data and reduce costs. These are enabled by double cluster sampling and external audit. PCS has been used for clinical research, assessment of the quality of radiotherapy and clinical QA. The national average is used as control data for radiotherapy in survey years. PCS has been conducted for various disease sites for more than 20 years and its reliability has been established in the USA. In the current study, PCS was performed only for the institutions of members of the cancer research groups by the Ministry of Health and Welfare in Japan (8-27 and 8-29) and their affiliated hospitals. Its feasibility and usefulness were clarified in our previous reports (4,5).

National averages can retrospectively and with a limited number of patients represent the actual state of medical care nationwide. Averages calculated by each stratum also show regional and institutional differences in the radiotherapeutic processes used for cancer patients. Comparison of these averages and data for individual institutions makes it possible to assess the quality of medical care in a particular institution. It should be noted that by making statistical adjustments to the data randomly sampled by using Sedransk's equation (3), the actual state of radiation oncology at the national level could be assessed with data for a limited number of patients.

For this study, a program for calculating national averages was prepared and an assessment of esophageal cancer treatment from 1992 to 1994 was simulated. Differences among institutional strata could be demonstrated for three aspects of esophageal cancer. These results indicate that the quality of medical care in A1 institutions was higher than that in B1 and B2 institutions.

Simple averages were higher than national averages calculated with Sedransk's equation (3). This is due to the higher ratio of A1 institutions sampled. Using Sedransk's equation, this overestimation of national averages could be prevented and the actual state of medical care could be assessed more accurately.

However, values of national average in this paper are not completely accurate. Instead of accurate values of the total number of institutions in each stratum, N_h, and the number of patients at each institution, M_hi, data from the structure survey by Tsunemoto in 1990 were used. It is necessary that we obtain accurate N_h and M_hi. This problem could be solved in new grant study (10-15) initiated by the Ministry of Health and Welfare (principal investigator: T.I.). Furthermore, the institutions sampled were not randomly selected, but were member institutions of the principal investigators (M.A. and H.I.) of the cancer research groups by the Ministry of Health and Welfare in Japan (8-27 and 8-29) and their affiliated hospitals. Therefore, there was bias that patients were not randomly selected. Moreover, because the institutions sampled in this study did not include A2 institutions, national averages in this study were calculated without taking into account A2 institutions. To obtain more accurate national averages, pure double random sampling will be necessary. Such pure double randomized sampling of institutions and their patients is also used for the new grant study (10-15).

Differences in the process among institutional strata could also be demonstrated in this study. This difference was found to be closely related to the institutional structure, including equipment and personnel, and should therefore have an effect on patient outcomes. For example, the average of application of brachytherapy was related to the capability of brachytherapy at each institution. Since there was the background that the ratio of A1 institutions that had brachytherapy equipment was higher than that of B1 and B2, average of A1 was higher than that of B1 and B2. In our series for this study, we have already reported that the survival of esophageal cancer patients who had not received surgery was influenced by the application of brachytherapy (10). Brachytherapy is not always utilized for the treatment of esophageal cancer patient; however, each institution needs to prepare the system for brachytherapy including equipment and personnel for essential cases. Although no B2 institution had brachytherapy equipment, 1.9% of patients were treated with brachytherapy in B2. These patients received brachytherapy at affiliated institutions. In PCS, tracing the information of such patients was easy by audits.

In summary, the program for calculating national averages using the methodology of PCS proved to be successful. A simulated calculation for esophageal cancer patients treated between 1992 and 1994 was performed and differences among institutional strata could be demonstrated. To increase the quality of care in radiation oncology in Japan, an accurate assessment of current radiotherapeutic processes on a nationwide basis is essential. National averages should prove to be a useful measure of QA in radiation oncology and also in other specialties in Japan

Acknowledgments

The authors thank all radiation oncologists (Table 1) who participated in this study. Their cooperation in providing information makes these surveys possible. This study was supported by grants from the Ministry of Education, Science and Culture (Overseas Research Fellowship), the Ministry of Health and Welfare (Grants-in-Aid for Cancer Research Nos 8-27, 8-29 and 10-15), the Japanese Society of Promotion of Sciences, Pfizer Health Research Foundation in Japan (96A014) and the Japanese Foundation for Multidisciplinary Treatment of Cancer. We sincerely thank Dr Lawrence R. Coia, Chairman of the Gastrointestinal Subcommittee of Patterns of Care Study, USA, for his great support and appropriate advice on performing audits for PCS in Japan.