The geography of hospital admission in a national health service with patient choice
研究了意大利每年约20%住院患者跨地区就医的现象，用引力模型和泊松伪最大似然法分析患者流动的影响因素，发现患者池规模对流出有抑制作用，但对吸引流入的影响因地区规模而异。
Each year about 20% of the 10 million hospital inpatients in Italy get admitted to hospitals outside the Local Health Authority of residence. In this paper we carefully explore this phenomenon and estimate gravity equations for 'trade' in hospital care using a Poisson pseudo-maximum likelihood method. Consistency of the PPML estimator is guaranteed under the null of independence provided that the conditional mean is correctly specified. In our case we find that patients' flows are affected by network autocorrelation. We correct for it by relying upon spatial filtering. Our results suggest that the gravity model is a good framework for explaining patient mobility in most of the examined diagnostic groups. We find that the ability to restrain patients' outflows increases with the size of the pool of enrollees. Moreover, the ability to attract patients' inflows is reduced by the size of pool of enrollees for all LHAs except for the very big LHAs. For LHAs in the top quintile of size of enrollees, the ability to attract inflows increases with the size of the pool.