39 studies were of addictive subtractive behaviour. These included: 35 studies of smoking (N = 24,627 subjects); and 4 studies of alcohol (N = 734 subjects).
17 studies were of non-addictive, subtractive/substitutive behaviour. These included: 12 studies of nutrition (N = 13,844 subjects); and 5 studies of weight (N = 198 subjects).
18 studies were of other behaviours. These included: 6 studies of contraception (N = 945 subjects); 3 studies of breast self- exam (N = 1502 subjects); 6 studies of injury prevention (N = 1669 subjects); 2 studies of stressor reduction (N = 164 subjects); and 1 study of exercise (N = 120 subjects).
Characteristics of primary studies are tabulated.
Overall the 74 studies were heterogeneous (Chi-squared = 621.5; P < 0.001).
The heterogeneity within behaviour groups is presented graphically.
Analysis by behaviour group: Addictive subtractive (smoking and alcohol): overall effect size = 0.61 (95%CI: 0.45, 0.77). Non-addictive, subtractive/substitutive (nutrition and weight): overall effect size = 0.51 (95%CI: 0.20, 0.82).
Other (contraception, breast-self exam, injury prevention, stressor reduction and exercise): overall effect size 0.56 (95%CI: 0.34, 0.77).
Effect sizes are presented separately for intervention and control variables by behaviour group. Higher scores on the Principles of Education Factor, Behaviour Change Support Factor and the use of self-monitoring were associated with larger effects. Having multiple sites was associated with smaller effects. Inconsistent directions across the behaviour groups were found for other intervention characteristics. Regression models for the three behaviour groups were homogeneous after removing outlier studies.
Details of the effects on the final sample of studies of applying inclusion criteria are reported elsewhere (see Other Publications of Related Interest).