Six types of dilution bias in population-based interventions
| Bias attributable to: | Definition |
| Exclusion of causes and consequences | |
| 1 Changes in non-intervening factors | Non-intervening factors in the community influence morbidity and mortality and distort the effect of an intervention. This is because of the non-randomised nature of community interventions. |
| 2 Single disease measurement | Most evaluations focus on a single disease measure although many behavioural lifestyle changes affect the risks of several diseases. |
| Exposure dilution | |
| 3 Population mobility | The fact that people move from the intervention area to the control area and vice versa will create a dilution bias, causing the effects to be underestimated in the intervention area and overestimated in the control area. |
| 4 Dissemination effects to other areas | Successful interventions are adopted by others, an effect omitted from the outcome analysis. |
| Mis-specification of follow up time | |
| 5 Social diffusion to following generations | The adult population exposed to the intervention influences the lifestyles of following generations, an effect usually omitted from the outcome analysis. |
| 6 Time lag | The effect of a risk factor reduction will have a lag time and be distributed during a long follow up time, which creates a dilution effect in the evaluation. |