Table 1 |
|
|
Hill's viewpoints or considerations for epidemiological causation |
|
| Consideration |
Explanation |
|
|
|
| Strength |
Strong associations are intuitively considered more compelling. However, weak associations
do not rule out causation. |
| Consistency |
The association is found in different experiments, with different populations, and
with varied circumstances. |
| Specificity |
The most controversial consideration. A cause leading to a single effect (and vice versa) offers more support for the causation argument than one cause leading to multiple
effects (and vice versa). |
| Temporality |
The cause must happen before the effect. |
| Biologic gradient |
A dose–response pattern is present, or incremental amounts of exposure should lead
to corresponding increments in the effect. |
| Plausibility |
The proposed association seems reasonable or probable as a cause. Most subjective
consideration. |
| Coherence |
A causative effect is not in conflict with current knowledge about the pathophysiology
of the disease. |
| Experimental evidence |
The effect can be reduced or altered by reducing or eliminating the proposed cause. |
| Analogy |
Alternative explanations for the causative effect are evaluated and considered less
likely than the one proposed. |
|
|
|
|
Gaffo et al. Arthritis Research & Therapy 2009 11:240 doi:10.1186/ar2761 |
|