Table 7.

Effectiveness of digital interventions on alcohol consumption, sorted by controls and further ordered by intervention type.

Review			Relevant studies, n (total studies)		Method of synthesis		Interventions		Outcomes		Follow-up		Summary of findings		AMSTAR-2 rating
Mixed (active and nonactive) controls
	Afshin et al (2016) [99]	47 (224)		Narrative synthesis of RCTsa and quasiexperimental studies		Various: internet, mobile, computer software, and sensors		Alcohol frequency and quantity, binge drinking, estimated blood alcohol concentration, alcohol dependency, and Alcohol Use Disorder Identification Test scores		1 week-2 years		34% (39/47) of studies, 41 RCTs and 6 quasiexperimental, reported a significant decrease in alcohol use; 83% (33/41) of RCTs reported statistically significant benefits.		Critically low
	Chebli et al (2016) [103]	2 (16)		Narrative synthesis of RCTs		Internet		Cessation and reduction of alcohol		1-12 months		Both studies demonstrated positive treatment outcomes in both arms, but there were no differences between internet intervention and control.		Critically low
	Kaner et al (2017) [77] (Alcohol only)	57 (57)		Narrative synthesis and meta-analysis of RCTs		Computer and mobile		Alcohol consumption and frequency		1-12 months		Alcohol consumption reduced by approximately 23 g per week (95% CI 15 to 30) at follow-up (1-12 months; based on 41 studies). Frequency of consumption reduced (based on 15 studies): participants who engaged with digital interventions had less than one drinking day per month fewer than no intervention controls (moderate‐quality evidence); had about one binge drinking session less per month in the intervention group (moderate‐quality evidence); and drank one unit per occasion less than no intervention control participants (moderate‐quality evidence). Compared with face-to-face interventions, there was no difference in alcohol consumption at the end of follow-up (mean difference 0.52 g/week; 95% CI −24.59 to 25.63; low‐quality evidence).		Low
	Kolar et al (2015) [78] (Alcohol only)	2 (18)		Narrative synthesis of all studies		Internet		Alcohol quantity and frequency		1 month		100% (2/2) of studies found reduced alcohol consumption in both arms but no significant differences between arms.		Critically low
	Palmer et al (2018) [111]	8 (71)		Narrative synthesis of RCTs		Mobile		Self-report alcohol consumption		Not reported		The effects of alcohol reduction interventions were inconclusive.		Moderate
	Rooke et al (2010) [112]	9 (34)		Meta-analysis of RCTs		Computer-delivered		Abstinence and reduction of alcohol		1-156 weeks		The weighted average effect size (Cohen d) was 0.20 (P<.001).		Critically low
	Vernon et al (2010) [81] (Alcohol only)	15 (15)		Narrative synthesis of all studies		Computer-delivered		Alcohol consumption		30 days-12 months		All but one intervention showed significant improvement in at least one drinking-related outcome. However, interventions were heterogenous and preintervention alcohol consumption was not standardized.		Critically low
	Webb et al (2010) [113]	9 (85)		Meta-analysis of RCTs		Internet		Alcohol consumption		Not reported		Small effects were observed for alcohol consumption (Cohen d+=0.14; k=9; 95% CI 0.00 to 0.27).		Critically low
Nonactive controls
	Black et al (2016) [82] (Alcohol only)	93 (93)		Meta-analysis of RCTs		Computer delivered		Alcohol consumption: total consumption over a period of time; average alcohol consumption per drinking occasion or drinking day; peak consumption—max consumed on one occasion. Frequency of heavy episodic drinking and of any alcohol consumption		Up to 2 years		Small effects averaging across timepoints, Cohen d=0.007 (heavy episodic drinking frequency) to Cohen d=0.15 (total consumption); in the short term, there were small-to-medium effects (Cohen d+=0.16 to 0.31) and significant effects on all outcomes except drinking frequency; in the medium-to-long term, they produced small (Cohen d+=0.07 to 0.12), significant effects on all outcomes.		Critically low
	Covolo et al (2017) [104]	1 (40)		Narrative synthesis of RCTs		Mobile		Alcohol frequency		0-2 years		Contrary to expectation, it was found that the mobile app significantly increased the frequency of drinking occasions compared with the control group (P=.001).		Critically low
	Riper et al (2011) [79] (Alcohol only)	9 (9)		Meta-analysis of RCTs		Internet		Alcohol consumption		Up to 12 months		An overall medium effect size (g=0.44; 95% CI 0.17 to 0.71; random effect model) in favor of the intervention groups was found.		Critically low
	Riper et al (2014) [80] (Alcohol only)	16 (16)		Meta-analysis of RCTs		Internet		Alcohol consumption		1 to 12 months		A small but significant overall effect size in favor of internet interventions (g=0.20; 95% CI 0.13 to 0.27; P=.001) was found. Participants in internet-based interventions consumed approximately 22 g of ethanol less and were more likely to adhere to low-risk drinking guidelines (risk difference 0.13; 95% CI 0.09 to 0.17; P=.001).		Critically low
	Tsoli et al (2018) [73]	4 (15)		Meta-analysis of RCTs		Interactive voice responses		Alcohol consumption		6 weeks-12 months		The meta-analysis of included studies demonstrated that interactive voice response–based interventions had no statistically significant effect on alcohol consumption (g=−0.077; 95% CI −0.162 to 0.007; k=4; P=.07).		Critically low

^aRCT: randomized controlled trial.