What Can We Learn From Large Databases? Lessons From Autoimmunity

The increasing use of computerized medical records has made the clinical data of the entire population available for epidemiological research. The resultant accessibility to this information mandates careful adaptions of ethical guidelines regarding the handling of clinical data. At the same time it grants a unique opportunity to explore the clinical nature of health and disease in large populations across all of society's strata, socioeconomic levels, ethnicities, and geographic locations regardless of their vicinity or distance to tertiary care centers. Analysis of large databases allows us to learn the public's behavior towards medical services and to investigate how medical interventions affect outcomes overtime. Moreover, the interaction between different co-morbidities can be better understood by large population studies. The huge numbers of patients involved in these studies provide a good model of multivariate analysis, a statistical tool that by following proper population adjustments underlines the true independent associations between different conditions. Nevertheless, the limitations of these studies should be borne in mind, such as in-built imprecision of diagnoses, incompleteness of the medical data, and the fact that these databases were initially planned for clinical and not investigational use.