Table 4.

Characteristics and vote counting of studies (n=7) not included in meta-analysis.

Study	DHIa type	Engagement measure	Physical activity measure	Association type	Association	Directionb
Carr et al [24]	Web-based	Activities completed	Steps	Multiple linear regression	Text only: “No other predictors,...[including engagement]...significantly explained...change in physical activity.”	0
Hansen et al [45]	Web-based	Logins	MVPAc	Kruskal-Wallis test	P≤.001	+
Hoj et al [46]	Smartphone app	Logins	Physical activity score	Multiple regression	SE −0.01 (0.067)	0
Maher et al [52]	Web-based	Logins	MVPA	Generalized linear mixed models	(F1,41=3.06; P=.04)	+
Marcus et al [22]	Web-based	Logins	MVPA	Quantile regression	β=34.32 (95% CI 14.33 to 54.31)	+
Rebar et al [54]	Web-based	Time	MVPA	Linear mixed models	γ=0.51 (95% CI −1.77 to 2.72); P>.05	0
Xian et al [56]	Smartphone app	Activities completed	Steps	Ordinal least squares regression	Every 10,000 XP points gained were associated with 2134 additional steps per day (95% CI 1673 to 2595; P<.001; R2=0.33])	+

^aDHI: digital health intervention.

^bA single association was selected for each study based on hierarchical criteria, see Methods section. Each study was summarized as either “+,” “−,” or “0.” “+” was assigned to the studies in which the point estimate and CI supported the hypothesis that higher engagement is associated with higher physical activity or reduced sedentary behavior. “0” was assigned to studies in which the point estimate and CI had inconclusive findings. “−“ was assigned to studies where the association point estimate and CI rejected the hypothesis. We assigned either “+” or “−“ to the studies without point estimates or CIs that reported significant association findings. We assigned “0” to the studies without point estimates or CIs that reported nonsignificant findings. See Multimedia Appendix 5 and Methods section for full details.

^cMVPA: moderate to vigorous physical activity.