1. |
- Agardh, EE, et al.
(författare)
-
The magnitude of bias in a cross-sectional study on lifestyle factors in relation to Type 2 diabetes
- 2006
-
Ingår i: Scandinavian journal of public health. - : SAGE Publications. - 1403-4948 .- 1651-1905. ; 34:6, s. 665-668
-
Tidskriftsartikel (refereegranskat)abstract
- Aim: In cross-sectional studies it may be difficult to ascertain the temporal order of exposure and disease, which may have consequences for causal inference. The authors aimed to illustrate the possible magnitude of this potential bias using data from a cross-sectional study on coffee consumption and work stress in relation to type 2 diabetes. Methods: By a series of computer simulations the authors examined to what extent the observed negative association between type 2 diabetes and high coffee consumption and positive association between type 2 diabetes and high work stress could be due to reverse causality, by assuming that cases changed their exposures in response to development of the disease. Results: If the negative association between coffee and type 2 diabetes was a consequence of reversed causality, 30—40% of the cases would have to decrease their coffee consumption from≥5 cups of coffee per day to 3—4 cups per day and from 3—4 cups per day to≤2 cups of coffee per day. Moreover, approximately 60% of the cases would have to increase their work stress from low to medium work stress and from medium to high work stress, in order to produce the positive association with diabetes that was observed. Conclusion: Even if the type 2 diabetic patients to some extent may have changed their exposure in response to disease development, it seems unlikely that the associations observed between type 2 diabetes, coffee consumption, and work stress are due to this bias.
|
|