Abstract
For the analysis of regional variations in hospital admission rates a model was set up, including both supply and demand indicators. Data were gathered for regions in four health care systems (the Dutch, Belgian, French and German), for 1982, 1979, 1974 (France 1982 only). Hospital bed supply proved to be the key factor in the explanation of the variation. The French health care system and, to a lesser extent, the German, turned out to be the strongest supply-dominated health care systems of the four. However, supply dominance in the German health care system seems to decrease over time. We concluded that the empirical generalisation, contained in Roemer's law, a bed built is a bed filled, is conditioned by time and place. Hypotheses on the conditional character of Roemer's law require further testing.