Few studies have been undertaken to assess the financial benefits of hospital information systems (HIS), particularly in the context of an overall portfolio type strategy. Moreover, the sometimes contradictory results of early works bring regularly the hospital decision makers' to make choices based primarily on the financial value of clinical or organizational benefits expected but not confirmed in the light of the high amounts invested. This work illustrates and discusses, from significant examples, results and limitations of studies to quantify the return on investment of HIS. By their very nature, accounting methods do not take into account the less tangible aspects of the impact of healthcare information technologies (HIT). Improved satisfaction of users (professionals and patients) is a typical example. The potential impact is clear, but still complex to translate financially. These methods also tend to underestimate certain aspects of the HIT capital. To achieve the best results these methods have to accumulate an exponentially increasing number of variables which might result in them failing as they become overly complex. In most cases, as the accuracy required increases, the amount of effort needed to feed the method rather than working the project also increases. Furthermore, all these financial methods, when used to evaluate a future investment, tend to be systematically biased against innovation. To compensate for these weaknesses, this work argues for the econometrics approaches. Econometrics is a branch of economics that has developed techniques to apply computational mathematics and statistics to the principles of the modern economy. The application of econometric methods in a company (microeconomics) is possible. In this context, the measurement of capital efficiency is achieved through the use of production functions or cost functions. In this work it is shown that: * The production functions are adaptable to the analysis of HIT in hospitals; * The impact of HIT is positive for the hospital production and that this impact is even greater if the HIT have a high level of integration; * It is possible to construct an accurate predictive model (0.61% precision) of this impact with existing investments; * This model can compute that the best proportion of HIT investment in hospitals should be around 10.6% of global investments. 10 If ICT investments have long been seen in hospitals as a necessary evil, accounting or econometric studies published in recent years suggests that it could be a necessary evil. The optimum of 10.6% found by an econometric model on data from AP-HP may be overstated if one accepts that the law of diminishing returns applies to the health sector as in other sectors economy. However, this is much higher than the figures recorded in France (1-2%) or USA (4-7%) and can be compared to those observed in strong HIS dependent sectors, such as banking or insurance (up to 16%). In advanced economies, HITs are a promising source of productivity growth, but they will have little direct contribution to the overall performance of a hospital if they are not combined with complementary investments in human capital, work practices, and some organizational restructuring. Finally, this work argues for the development of a European observatory of the level of computerization in hospitals, which represents the only way to measure the medium and long term impact of HIT investments that will necessarily be made in the sector.
