[Statistical models and evaluation of observation studies].

Although randomized clinical trials can be regarded as the gold standard in many fields of clinical research, non-randomized (so called observational) studies are indispensable. Methodologic quality of such studies is at least as important as that of randomized trials but has been much less standardized. The analysis and interpretation of the results of observational studies is a very demanding job which needs a high level of knowledge and experience--medical as well as statistical. Basis for these analyses are complex statistical models the grounds and application of which are discussed in the following article.