The present study was an investigation of variations in cancer surviva
l rates among hospitals in Japan, focusing in particular on the number
of hospital beds as an institutional characteristic. Using data from
11 population-based cancer registries, the three-year survival rates f
or stomach cancer (n 1665), colorectal cancer (n 1090) and lung cancer
(n 895) patients diagnosed in 1985 were calculated according to three
different hospital categories (100-299, 300-499, 500+ beds). Cox's pr
oportional hazards model was conducted, with adjustments for sex, age,
clinical stage at diagnosis and treatment status, excluding patients
who had been detected by screening (asymptomatic cases). The stomach a
nd lung cancer patients treated in small hospitals (100-299 beds) were
at a significantly higher risk of death than those treated in large h
ospitals (500+ beds) (hazard ratio (HR)=1.36, 95% confidence interval
(CI)=1.11-1.65; HR=1.41, 95% CI=1.13-1.77, respectively). Similar find
ings were observed among colorectal cancer patients although they were
not statistically significant. The findings can provide some informat
ion useful for the development of future public health policies aimed
at controlling cancer mortality rates in our country.