Purpose: We examined whether the data registered in the Japanese Patterns of Care Study (PCS) for patients with non-small-cell lung cancer (NSCLC) represent the actual situation of radiotherapy in Japan. The Osaka Cancer Registry (OCR) data, forming the largest database of a regional cancer registry in Japan, were adopted for use as a benchmark against the national condition.
Patients and methods: We examined 906 patients of the PCS treated between 1995 and 1997 and 845 patients of the OCR registered between 1988 and 1992. The investigation was made by descriptive statistical methods to measure age, stage, combined treatments, type of treated hospitals, and prognosis. Furthermore, the national averages (NAs) of the PCS process (PCS NA) were also calculated to compensate for the imbalance in the PCS data sampling.
Results: The mean age was 67.3 +/- 10.1 in PCS and 64.4 +/- 11.0 in OCR (p < 0.001), 67.2 in PCS NA. The male ratio was 84.2% in PCS and 84.0% in OCR (p = 0.411), 84.1% in PCS NA. The ratio of the patients at the localized stage was 24.2% in PCS and 15.6% in OCR (p = 0.001), 21.1% in PCS NA. The ratio of surgery combined was 24.2% in PCS and 28.9% in OCR (p = 0.026), 25.3% in PCS NA. The ratio of chemotherapy combined was 50.1% in PCS and 67.5% in OCR (p = 0.001), 47.4% in PCS NA. Because the definitions of institution classification and period of prognostic inquiry were different between the two databases, the 3-year survival rates were calculated for reference. In the nonsurgery group, it was 20.3% in PCS and 11.3% in OCR (p = 0.001), and in the surgery group it was 52.5% in PCS and 42.2% in OCR (p = 0.057).
Results: Ages in the two databases were inconsistent. Sex distributions were consistent. Surgery and chemotherapy were more frequently performed for the OCR patients, and more patients at more advanced stages were also observed in OCR. The PCS NAs of sex, stage, and ratio of surgery combined were at the midpoints between those of PCS and OCR. The survival rate of NSCLC patients in the OCR was significantly inferior to that in the PCS. The follow-up rate of the PCS was lower than that of the OCR. The general features of PCS data showed similarity to OCR data, and the results of the PCS NAs suggested the effectiveness of this method to adjust the sampling imbalance in PCS.