Smartphone-based alcohol interventions: A systematic review on the role of notifications in changing behaviors toward alcohol

Figures & data

Figure 1. PRISMA Flow diagram of search strategy.

Table 1. Characteristics of included studies.

Reference	Study			Sample		Intervention			Notifications				Outcomes
Lead Author, Year	Location	Design	Aim	Sample Size, Gender (% female), Age	Population/ Occupation	Name, Mode of Delivery	Duration	Description/ Aim of Intervention	Mode of Delivery, Frequency	Personalization	Development	Content	Outcomes
Attwood, 2017	UK	Cohort, mixed-methods	To evaluate an existing publically available app, designed by Drinkaware, a UK-based alcohol awareness charity.	Quantitative: 119,713 downloaded and entered data on app. Qualitative: 189 completed survey, 40 interviewed. Age: 18–65 years	General population.	Drinkaware Intervention delivered via smartphone app.	Not reported – data for this study collected over 13 months (August 2014 – September 2015).	To enable users to: (1) calculate the units and calories in drinks and to track alcohol consumption over time, (2) gain feedback on how drinking impacts health and to understand trends in drinking patterns, (3) set goals to reduce drinking that are relevant to their lifestyle and to receive supportive notifications when specific achievements are met, and (4) define geographic locations where users may feel that additional support to regulate alcohol consumption is needed.	Supportive notifications. GPS alerts in high-risk locations.	No mention of personalization. Interviews suggested needs more personalized content.	Not reported.	The app sends users supportive messages when they reach their “weak spot” locations (e.g. “You are near one of your designated weak spots. Remember, drinking less has many feel-good benefits”).	To monitor and reduce alcohol consumption.
Bell, 2020	UK	Longitudinal, observational	To visualize behavioral engagement with Drink Less, a behavior change app to help reduce hazardous and harmful alcohol consumption in the general adult population of the United Kingdom.	N = 19,233 50.4% female Age: 18+ years (m = 44)	General population.	Drink Less Smartphone application.	Data for this study from users who downloaded app between May 2017 – January 2019. Minimum 84 days use measured.	To help its users reduce hazardous and harmful alcohol consumption.	Daily notifications sent at 11am.	No mention of personalization.	Not reported.	The app sends a local daily push notification at 11 AM that asks users to “Please complete your drink diaries,” to encourage self-monitoring of drinking behavior.	To explore the association between the delivery of the notification and subsequent near-term engagement of opening the app.
Dulin, 2014	USA	Pilot, cohort	To assess early-stage effectiveness and usability of a smartphone-based intervention system that provides a stand-alone, self-administered intervention option, the Location-Based Monitoring and Intervention for Alcohol Use Disorders (LBMI-A).	N = 28 46% female Age: 22–45 years	Clinical population – met criteria for alcohol use disorder (but not having treatment).	LBMI-A A smartphone system.	6 weeks.	The Buddy Steps included: (1) enhancement of motivation for change by providing assessment feedback; (2) high-risk for drinking locations and strategies for avoiding them; (3) importance of supportive people; (4) managing alcohol cravings; (5) managing life problems through problem solving strategies; (6) pleasurable non-drinking activities; and 7) productive communication strategies that emphasise assertive communication.	GPS alert when in high risk location.	No mention of personalization.	Not reported.	After high risk locations were entered, the system utilized the GPS capabilities of the smartphone to provide an audible alert and suggestions for maintaining control of drinking when they crossed a boundary around a specific high risk location.	To assess the effectiveness and usability of the smartphone-based system to reduce alcohol consumption for those with AUDs.
Dulin, 2017	USA	Pilot, cohort	To compare data generated from a smartphone-based, daily alcohol assessment methodology with results from the TLFB amongst participants who met criteria for DSM-5 criteria for alcohol use disorder who were utilizing a smartphone intervention app to change their drinking on commonly utilized alcohol outcome variables.	N = 25 48% female Age: 22–65 years (m = 33.6)	Clinical population – met DSM-5 criteria for AUD and be at least minimally motivated to change drinking behavior.	LBMI-A A smartphone system.	6 weeks.	The LBMI-A provided a stepwise progression through seven treatment modules designed to enhance motivation to change, improve social support for sobriety, develop awareness of alcohol triggers, and improve coping methods.	Smartphone prompts to remind to log drinking.	No mention of personalized messages.	Not reported.	Each day at a pre-set time, participants received a smartphone prompt to record the number of standard drinks consumed during the previous 24 hours.	To compare the consistency of alcohol consumption data collected through prospective and retrospective methods.
Garnett, 2019	UK	App development, qualitative	To describe the development of a smartphone app to help individuals who drink excessively to reduce their alcohol consumption.	Not reported.	General population.	Drink Less Smartphone application.	Not reported.	Drink Less is an app for the general population of adults seeking digital support to reduce excessive alcohol consumption.	Alert, prompts, in-app notifications.	No mention of personalized messages.	Not reported.	Users were reminded to complete a log of their drinking at 11:05 am each morning, primarily through an on-screen alert.	To develop the Drink Less app through: (1) selecting intervention components, and (2) designing and translating the components into an app.
*Gustafson, 2014	USA	RCT	To determine whether patients leaving residential treatment for AUDs with a smartphone application to support recovery have fewer risky drinking days than control-group patients.	N = 349 39% female Age: 18+ years (m = 38)	Clinical population – met the criteria for DSM-IV alcohol dependence when entering residential treatment.	A-CHESS Smartphone application.	Intervention: 8 months with 4-month follow-up. Control: 12 months treatment as usual.	A-CHESS had both static content (e.g., audio-guided relaxation) and interactive features. For example, if a patient neared a high-risk location (a bar she used to frequent), GPS initiated an alert asking the patient if she wanted to be there.	GPS initiated alerts when in ‘high risk’ location.	No mention of personalized messages.	Not reported.	If a patient neared a high-risk location, GPS initiated an alert asking the patient if they wanted to be there.	To report on risky drinking days in the previous 30 days using surveys taken 4, 8, and 12 months after discharge from residential treatment.
Malte, 2021	USA	Pilot, cohort	To evaluate the usability and acceptability of Step Away, a mobile app designed to reduce alcohol-related risks, and explored pre-post changes on alcohol use, psychological distress, and quality of life.	N = 55 16% female Age: 18–55 years (m = 37.5)	Clinical population – screened positive for alcohol misuse.	Step Away Smartphone application.	6 months.	Step Away, the next iteration of LBMI-A, is an iOS-based mobile app designed to help people self-manage drinking and alcohol-related problems and thus is an alternative for those who do not want or have access to traditional treatment.	Daily push notifications – including prompting users to complete mood questionnaire, drinking behaviors, alcohol cravings and alcohol triggers.	Users can personalize the app through the use of reminders about reasons for changes, high-risk times, and scheduled activities.	Not reported.	Using push notifications, Step Away prompts users daily to complete a brief self-monitoring questionnaire on current mood, drinking behaviors, and alcohol cravings and triggers for alcohol use during the prior 24 hours.	To evaluate the usability and acceptability of Step Away. To explore pre-post changes on alcohol use, psychological distress and quality of life.
*McTavish, 2012	USA	RCT	To report on the relative impact and use of A-CHESS four months after patients entered the study and discusses implications of the results on treating addiction and chronic diseases generally.	N = 349 39.4% female Age: m = 28.3 years	Clinical population – alcohol dependence leaving residential treatment.	A-CHESS Smartphone application.	Intervention: A-CHESS plus TAU. Intervention for 8 months, but followed for 12 months. Control: TAU.	A-CHESS was designed for people in recovery being discharged from residential care, and has digital voice services, text messaging, Web access, GPS, voice recognition, and video capabilities.	Panic button leads to automated reminders. GPS initiated alerts.	No mention of personalized messages.	Not reported.	Global positioning system (GPS) technology tracks when participants approach an area where they traditionally obtained or consumed alcohol so they can receive “just-in-time” support to work through what might be a high-risk situation for relapse. To activate, individuals voluntarily register places where they regularly obtained or consumed alcohol in the past and now designate as a high-risk locations for relapse.	To explore the impact of smartphone based system (A-CHESS) and implications for the treatment of alcohol dependency after leaving residential treatment.
Milward, 2018	UK	Qualitative	To (1) understand why and how participants engaged with the BRANCH app, (2) explore facilitators and barriers to engagement with app features, (3) explore how the BRANCH app impacted drinking behavior, (4) use these data to identify typologies of users of the BRANCH app in terms of engagement behaviors, and (5) identify future eSBI app design implications.	N = 20 80% female Age: 18–30 years (m = 24)	Not reported.	BRANCH Web-based application.	28 days	The comprehensive (intervention) version included a Twitter or Facebook style newsfeed enabling interaction between app users, as well as providing tailored notifications, motivational messaging (including positive reinforcement and praise), and in-app reminders based on goals.	Tailored notifications and in app reminders (on smartphone).	Yes.	Not reported.	BRANCH was tailored; when signing up for the app, users selected their motivations for cutting down drinking. Personalized feedback and tailored information were delivered to users based on their selection of motivators.	To explore distinct types of engagers, complex typologies were developed involving the interconnection between dimensions.
Monk, 2015	UK	Cohort	To compare real-time measurements of alcohol consumption with retrospective accounts of alcohol consumption to examine possible discrepancies between, and contextual influences on, the different accounts.	N = 69 41% female Age: 18–36 years (m = 21.47)	General population – recruited on university campus.	Smartphone app designed specifically for study.	Not reported.	A smartphone application was designed specifically for this research and it enabled participants to respond to questioning via the use of their own mobile phone.	Hourly participation prompts.	No mention of personalized messages.	Not reported.	It provided hourly prompts that asked respondents to record what they were drinking.	To compare real time versus retrospective alcohol consumption.
Muroff, 2017	USA	Cohort, pilot	To examine the feasibility and acceptability of delivering the theory-informed smartphone relapse prevention system CASA-CHESS to Spanish-speaking Latinos who successfully completed residential treatment and analyzes the specific usage data from this recovery tool.	N = 79 11% female Age: 18+ years (m = 41.1)	Clinical population – Spanish- residential treatment program for alcohol and other drug disorders/mental disorders.	CASA-CHESS Android smartphone app.	4 months.	The smartphone relapse prevention tool CASA-CHESS was developed by adapting and translating A-CHESS to be linguistically and culturally relevant to Spanish-speaking Latinos in recovery. CASA-CHESS replicated aspects of the original design of A-CHESS, providing “digital voice services, text messaging, web access, GPS, voice recognition, and video capabilities”.	Sent automated medication adherence surveys daily.	No mention of personalized messages.	Not reported.	Not reported.	To assess the feasibility and acceptability of alcohol recovery app for Latinos (with comorbid drug and mental health disorders) – app usage over 4 months.
**Poulton, 2018	Australia	Cohort	To examine differences between alcohol consumption information captured using an app for 21 days with data recorded via traditional, well-validated retrospective measures – namely, 21-day TLFB and the AUQ – in a large healthy sample.	N = 671 70% female Age: 16–56 years (m = 23.1)	General population – healthy individuals.	CNLab-A Smartphone application.	21 days.	To assess alcohol intake behavior in real-time.	Notification prompts, twice daily.	No mention of personalized messages.	Not reported.	One notification is pre-set to 8 am while the other can be set to suit the user. While participants are directed at the outset to record alcohol consumption as it happens (or as soon thereafter as possible), notifications serve to prompt individuals to input information twice daily in case they neglect to do so when drinking.	To compare real time versus retrospective alcohol consumption.
**Poulton, 2019	Australia	App development, cohort	To describe the process used to guide the development of a custom-built smartphone app designed to capture alcohol intake behavior in the healthy population.	N = 671 70% female Age: 16–56 years (m = 23.1)	General population.	CNLab-A Smartphone application.	21 days.	To assess alcohol intake behavior in real-time.	Notification prompts, twice daily.	No mention of personalized messages.	Not reported.	One notification is pre-set to 8 am while the other can be set to suit the user. While participants are directed at the outset to record alcohol consumption as it happens (or as soon thereafter as possible), notifications serve to prompt individuals to input information twice daily in case they neglect to do so when drinking.	To describe the process used to guide the development of a custom-built smartphone app designed to capture alcohol intake behavior in the healthy population for research purposes. To evaluate methodological issues related to compliance with and reactivity to study protocols as a function of hazard and non-hazard drinker status.
Smith, 2017	Australia	App development, mixed-methods	The AlcoRisk app for studying alcohol consumption and risk taking tendencies is presented alongside a software development process that integrates these multidisciplinary fields.	N = 19 42% female	General population – recruited through University of Tasmania.	AlcoRisk App EMA smartphone app.	2 weeks.	The AlcoRisk app is one such EMA smartphone app developed to study alcohol consumption and associated risk taking tendencies at the time of occurrence within natural environments.	Automated push notifications and prompts.	None. The AlcoRisk app's personal utility was reported as low with no personal feedback on alcohol consumption.	Not reported.	Morning reports were prompted at 10:00 am via a smartphone device notification to enter alcoholic beverages consumed along with their current affective and physical state. A reminder notification was also issued when research participants had unfinished drink entries. A reminder notification was issued when research participants closed an unfinished assessment.	To explore compliance, reactivity and usability of the app.

Notes. *Studies use the same data set. **Studies use the same data set.

Table 2. Quality assessment scores.

Reference	Exposed cohort representative of target population (0–1)	Exposed and non–exposed groups drawn from the same cohort  (0–1)	Report on the use of notifications, app engagement or usage (0–1)	Sample size justified and satisfactory (0–1)	Controls for factors (0–1) either in selection of cohort or adjusted/stratified analysis	Follow–up ≥3 months  (0–1)	Follow–up complete, or unlikely to introduce bias (<20% lost or attrition described and accounted for in analysis)  (0–1)	Statistical test reporting clear, appropriate and complete  (0–1)	Total score (0–8)
Gustafson, 2014	1	1	1	1	1	1	0	1	7
McTavish, 2012	1	1	1	1	1	1	0	1	7
Attwood, 2017	1	1	1	1	0	1	0	1	6
Bell, 2020	1	N/A	1	1	1	1	0	1	6
Dulin, 2014	1	N/A	1	0	0	0	1	1	4
Dulin, 2017	1	N/A	1	0	0	0	1	1	4
Garnett, 2019	1	N/A	1	N/A	N/A	0	1	1	4
Malte, 2021	1	N/A	1	1	0	1	0	1	5
Milward, 2018	1	1	1	1	1	0	1	1	7
Monk, 2015	1	N/A	1	0	0	0	0	1	3
Muroff, 2017	1	N/A	1	0	0	1	0	1	4
Poulton, 2018	1	N/A	1	1	0	0	1	1	5
Poulton, 2019	1	1	1	1	1	0	1	1	7
Smith, 2017	1	N/A	1	1	0	0	0	0	3

Note. Adapted version of Newcastle-Ottowa Scale (NOS). Dashed line separates RCTs and non-randomized studies.