Etiologic Studies’ Essentials

  • O. S. Miettinen
  • I. Karp


Given that a defined study base is a sine-qua-non for any admissible type of etiogenetic study, and that each person-moment in it is to represent the domain of the designed model for the cases’ rate of occurrence, it generally cannot be operationally formed, nor even directly defined; it must be defined as a segment of the population-time within a defined source population-time, a source base. The latter, in turn, is formed by the course of a defined source population over a particular span of time, with this population possibly defined indirectly, as the catchment population (of cases) secondary to the directly-defined manner of case identification.

Cases of the outcome event are identified, comprehensively, in the source-base experience, and a fair sample of the source base (of the infinite number of person-moments constituting it) is drawn. The resulting first-stage case and base series are reduced to instances from the actual study base (as defined), to the actual pair of study series.

The model – log-linear – for the cases’ rate of occurrence implies its corresponding logistic model to be fitted to the data (with Y=1 and Y=0 for the case and base series, respectively), yielding a result for the rate ratio of interest, as a function of the modifiers of this as they are accounted for in the model. This, in turn, provides for calculation of the ‘expected’ number of index cases in the calculation of the essential result, that for the factor-conditional etiogenetic proportion.

These calculations under the designed model lack the intuitive appeal of those based on cross-stratification of the data according to the confounders being controlled. A solution to this problem is stratification by a unidimensional confounder score, provided by the fitted logistic function evaluated at the reference category of the etiogenetic determinant under study. Examination of the data in these strata allows verification of freedom from confounding by the extraneous determinants accounted for, and the calculation of the desired result is intuitive and straightforward.


Study Base Source Population Base Series Calendar Time Source Base 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Copyright information

© Springer Science+Business Media Dordrecht 2012

Authors and Affiliations

  • O. S. Miettinen
    • 1
    • 2
  • I. Karp
    • 3
  1. 1.McGill University Université de MontréalMontréalCanada
  2. 2.Cornell UniversityNew YorkUSA
  3. 3.Université de MontréalMontréalCanada

Personalised recommendations