[Big data, Roemer's law and avoidable hospital admissions].

From an analysis of data from 23 European countries to determine the impact of primary care on avoidable hospital admissions for uncontrolled diabetes it appeared that, contrary to expectation, countries with strong primary care did not have a lower rate of avoidable hospital admission. It is clear that Roemer's law, 'a bed built is a bed filled,' still applies. However, the validity of this sort of analysis can be questioned, as these data are highly aggregated, and registration quality differs between countries. It is also questionable if these datasets can be considered as 'big data' as there are relatively small numbers per country. Big data analyses are useful for discerning patterns and formulating hypotheses, but not for proving causality. An unwanted side effect of this kind of analysis might be that policymakers use these not so valid results to underpin their policy to their advantage.