Fractal meta-analysis and 'causality' embedded in complexity: advanced understanding of disease etiology.

Epidemiologists have gradually come to realize that repeated testing of 'causal' hypotheses normally results in inconsistent outcomes. Contradictory results pop up often enough to arouse suspicion that central epidemiological paradigms may be invalid. In this paper, we introduce complexity to explain inconsistency. Linear models produce inconsistency simply because they are not sufficiently rooted in the complex nature of living things. We design a meta-analysis consisting of (a) fractal investigation, used in this paper to reveal the possibly fractal nature of ordered series of relative risks (RR) and their possible Self-Organized Criticality (SOC); (b) disclosure of complexity-bound associations (DCBA) analysis which is used to disclose law-like and chaos-like associations of exposure and risk of disease within the series. We use (a) and (b) to reanalyze three published meta-studies, one of which investigates the possible association of oral contraceptives and female breast cancer. We demonstrate that the OR-series of the latter is fractal and in a state of SOC and conclude, contrary to the authors of the original meta-study, that there is no law-like association of oral contraceptives and breast cancer in this series. We achieve similar results for the other two meta-studies, results that are highly relevant for the clinical recommendations given by the authors.