Table 1

Six types of dilution bias in population-based interventions

Bias attributable to: Definition
Exclusion of causes and consequences
1 Changes in non-intervening factorsNon-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 measurementMost evaluations focus on a single disease measure although many behavioural   lifestyle changes affect the risks of several diseases.
Exposure dilution
3 Population mobilityThe 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 areasSuccessful interventions are adopted by others, an effect omitted from the  outcome analysis.
Mis-specification of follow up time
5 Social diffusion to following generationsThe adult population exposed to the intervention influences the lifestyles of  following generations, an effect usually omitted from the outcome analysis.
6 Time lagThe 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.