From trials to communities: implementation and scale-up of health behaviour interventions

Table 2 Results of the logistic regression examining trial characteristics associations to scale-up (n = 49)

Trial characteristics	Number of interventions n = 49 (%)	Unadjusted		Adjusted model^c
Trial characteristics	Number of interventions n = 49 (%)	Odds ratio (95% CI)	p-value	Odds ratio (95% CI)	p-value
Intervention effectiveness
No effect	5 (45%)	Ref.	0.50	Ref.	0.69
Potentially effective only^a	8 (38%)	0.74 (0.17, 3.24)		0.72 (0.14, 3.81)
Effective^b	20 (29%)	0.50 (0.14, 1.83)		0.55 (0.12, 2.45)
Risk of bias
Unclear + High risk	22 (34%)	Ref.	0.70	Ref.	0.39
Low risk	11 (31%)	0.84 (0.35, 2.02)		0.64 (0.23, 1.78)
Settings
Community + worksite	6 (27%)	Ref.	0.81	Ref.	0.52
Education	23 (34%)	1.39 (0.48, 4.04)		1.13 (0.33, 3.83)
Medical + other	4 (36%)	1.52 (0.32, 7.15)		2.54 (0.46, 14.20)
Involvement of end-users (n = 48)^d
Not at all + a little	13 (27%)	Ref.	0.23	Ref.	0.83
Substantially	20 (38%)	1.68 (0.72, 3.92)		1.11 (0.43, 2.89)
Dissemination strategies mean number (SD)	7.12 (2.04)	1.48 (1.19, 1.83)	< 0.001	1.50 (1.19, 1.88)	< 0.001

^aTrials were deemed potentially beneficial if the effect of the intervention on secondary outcomes was significant (p < 0.05), but the effect on the primary outcome was not significant
^bTrials were deemed effective if the effect of the intervention on the primary outcome was significant (p < 0.05)
^cModel adjusted for intervention effective, risk of bias, setting, involvement of end-user, and dissemination strategies
^dMissing data for one respondent who preferred not to answer this question
Boldface indicates statistical significance

ISSN: 1478-4505