Administrative data from 2175 gastric cancer patients was analyzed using two Cox proportional hazard models with (i) delay in diagnosis and (ii) survival as dependent variables. Density of general practitioners, density of gastroenterologists, characteristics of specialty treatment centers, demographic information, and comorbidities were included in the models. Differentiation was made between urban and rural areas.
The likelihood of being diagnosed increased with an increase in general practitioners (p < 0.0001) and gastroenterologists (p < 0.0001) in rural areas. In urban areas a higher density of general practitioners reduced delay in diagnosis (p = 0.0262), while a higher density of gastroenterologists did not (p = 0.2480). The number of gastric cancer cases performed in hospital had a positive impact on survival (p < 0.0001), while outpatient infrastructure did not.
Delay in diagnosis can be reduced by higher availability of general practitioners and gastroenterologists in rural areas. Given the already very high density of physicians in urban areas there is no effect of additional gastroenterologists. As learning effects can be observed with increased hospital volumes, minimum volumes for treatment of gastric cancer may be defined.