Mobile health (mHealth) usually refers to smartphone-based medical support for patients. mHealth can reinforce treatment from behavioral health professionals by offering patients evidence-based care at the moment they need it. The Addiction-Comprehensive Health Enhancement Support System (A-CHESS) is a multicomponent smartphone application designed to provide continuing support to people in substance use recovery (Gustafson et al., 2014). With a design consistent with self-determination theory (Ryan & Deci, 2000), A-CHESS seeks to increase patients’ coping competence by strengthening skills that prevent relapse, enhancing social relatedness by connecting users to sources of social support, and promoting autonomy in managing recovery from addiction. The A-CHESS mHealth system aims to help individuals gain knowledge and skills, connect with others who share their struggles, and set their own goals. A-CHESS users can choose to engage with some or all of its components.
In a randomized controlled trial, individuals assigned to receive A-CHESS had significantly fewer risky drinking days and increased abstinence over twelve months than those in the control condition without A-CHESS. The trial involved 349 participants with alcohol use disorders (AUDs) who were discharged from nonprofit residential treatment programs. Participants were randomly assigned to A-CHESS for eight months or a control condition that received only the survey assessments given to all participants (Gustafson et al., 2014). The trial findings raised questions about how A-CHESS reduced drinking. Identifying how mHealth platforms reduce drinking would help clinicians, administrators, and researchers understand how to better design and use these innovative treatments to support patients with addiction disorders. Our group did additional analysis on the A-CHESS trial data to investigate how this mHealth resource supported patients in recovery from AUD.
Some health researchers propose that mHealth interventions are best used to strengthen connections between patients and their health care systems, rather than relying on mHealth solutions to produce change on their own via psychological mechanisms (e.g., by increasing self-efficacy; Mohr, Burns, Schueller, Clarke, & Klinkman, 2013). Thus, technology-based, recovery support interventions such as A-CHESS may encourage people to seek additional in-person treatment. We hypothesized that, in the A-CHESS trial that was conducted among individuals discharging from residential addiction treatment, recovery support features in A-CHESS influenced patients to seek aftercare.
Aftercare, also known as “step-down care” or “continuing care,” refers to addiction treatment received after an initial intensive treatment phase such as residential care. Aftercare may take the form of outpatient treatment or mutual-help programs such as Alcoholics Anonymous (AA) or Narcotics Anonymous (NA; McKay, 2005, 2009). This new phase of treatment may be initiated to reduce risk of relapse and/or to maintain progress after a relapse occurs. Given that substance use disorders (SUDs) can manifest as chronic conditions, keeping patients engaged in treatment systems is often a desirable goal (McKay, 2005; McLellan, Lewis, O’Brien, & Kleber, 2000), and systematic reviews suggest that aftercare can improve addiction-related outcomes.
Thus, if mHealth recovery-support interventions such as A-CHESS facilitate the use of outpatient addiction treatment or mutual-help programs following residential care, they could be an effective approach to improving patient outcomes. To our knowledge, no studies have examined whether mHealth interventions increase the use of services for addiction following residential care. We also do not know how use of additional services promoted by an mHealth intervention affects substance use outcomes. Understanding these associations for A-CHESS would help explain the mechanisms of behavior change by which the mHealth resource reduced risky drinking and increased abstinence.
The study reported here was a new analysis of data from the A-CHESS randomized controlled trial. We examined the extent to which A-CHESS increased utilization of postdischarge addiction services, specifically outpatient addiction treatment and mutual help. We examined the extent to which use of these addiction services was responsible for the effects of A-CHESS on risky drinking days and abstinence. We hypothesized the following:
- Participants assigned to the A-CHESS study arm used more postdischarge addiction services in the year following discharge, compared to controls
- Use of postdischarge addiction services was associated with reduced drinking days and increased abstinence
- Use of postdischarge addiction services specifically attributable to A-CHESS helped reduce risky drinking days and abstinence
The A-CHESS randomized controlled trial recruited participants from five residential addiction treatment programs in two addiction treatment organizations two weeks before their anticipated date of discharge. To be recruited into the study, patients had to have DSM-IV alcohol dependence upon entering residential treatment, be age eighteen or older, be willing to be randomized into a treatment or a control condition, and identify two contacts to help reach them. We excluded patients who had significant developmental impairment, cognitive impairment, or vision problems that would limit the ability to use the smartphone application. A total of 349 adults with AUDs were enrolled (91.8 percent of those approached). A more detailed description of the participants and procedures is in Gustafson et al., 2014.
Participants were randomized to receive the A-CHESS intervention for eight months or to receive treatment as usual. Intervention group participants were provided with a smartphone and an eight-month service plan. The smartphone was loaded with the A-CHESS application and participants were taught to use it before discharge under counselor guidance. After this brief training, intervention-arm participants were required to demonstrate a minimal understanding of smartphone use, text messaging, and basic A-CHESS use; set up an anonymous profile; and load contact information for two people to receive text messages if patients pressed a panic button in the application. A research team member served as an “A-CHESS coach,” providing intervention group participants with regular, empathic outreach, primarily through electronic discussion board posts. Intervention arm participants kept the smartphone at the end of the eight-month intervention period and could continue using the A-CHESS system, if they desired, by paying for their own service plan or using wireless Internet. Approximately 80 percent of participants assigned to the A-CHESS arm continued to access the system four months after randomization (McTavish, Chih, Shah, & Gustafson, 2012), and 57.6 percent of participants used A-CHESS at least once in the last week of the eight-month intervention (Gustafson et al., 2014). Intervention design, recruitment, and follow-up procedures were described previously (Gustafson et al., 2014; McTavish et al., 2012).
Primary and Secondary Outcomes
All participants were followed for one year after randomization and surveys were administered by the research team in person at baseline and by telephone at four, eight, and twelve months. The primary outcome of the A-CHESS trial was risky drinking days assessed by survey, using the National Institute on Alcohol Abuse and Alcoholism (NIAAA) definition of a standard drink. One item asked patients to report the number of risky drinking days in the previous thirty days, defined as exceeding four standard drinks in a two-hour period for men or three standard drinks in a two-hour period for women. Abstinence was a secondary outcome, assessed with one item, which we coded as positive for those who reported consuming zero drinks in the previous thirty days.
Postdischarge Outpatient Treatment and Mutual-Help Program Attendance
We assessed postdischarge outpatient addiction treatment services by asking participants, “Have you been back to treatment in the past month?” followed by the question, “Are you going to outpatient treatment?” Assessment was at months four, eight, and twelve. To assess mutual-help attendance, at each follow-up survey we used an item from the Brief Addiction Monitor (Cacciola et al., 2013; Nelson, Young, & Chapman, 2014) that assesses past-week AA or NA group attendance: “How many days did you attend self-help meetings like AA or NA to support your recovery?”
Patient Characteristics Used in Adjustments
We measured several variables to statistically adjust for differences in patients at baseline. Patient interviews assessed age, gender, race, mental health problems beyond SUDs, and lifetime use of addiction treatment prior to entering residential care, since these variables may influence both treatment utilization and study outcomes (Glass et al., 2010; Glass, Grant, Yoon, & Bucholz, 2015). As a measure of motivation for treatment, we assessed reasons for entering residential care (e.g., treatment sought per own initiative, family/friend pressures, employer pressures, court referral, or family services referral).
We described the sample by calculating the mean and standard deviation of continuous variables. For variables represented by categories, we calculated the total number and proportion of people in each category.
Differences in Outcomes
To estimate differences in the outcomes between the A-CHESS and control groups, we used statistical models appropriate for measuring count variables (i.e., risky drinking days) and categorical variables (i.e., postdischarge outpatient addiction treatment, mutual help, and abstinence). These analyses accounted for all participant characteristics described above. We also accounted for having multiple observations of the same participant over time, as well as for having participants at multiple sites.
Conceptually, our mediation analysis sought to explain the extent to which A-CHESS increased the use of postdischarge outpatient addiction treatment and mutual-help programs, and how much these changes in service utilization attributed to A-CHESS produced better outcomes for risky drinking days and abstinence. Modern statistical methods can estimate an “indirect effect” that quantifies the extent to which a change in an independent variable affects a change in the dependent variable specifically through its influence on a mediating variable (MacKinnon, 2008). Similar to our main analysis, we used appropriate procedures for modeling the outcome variables, and adjusted for participant characteristics and multiple measurements over time.
To describe these effects in ways that are meaningful to researchers and health professionals, we calculated percent change in the expected number of risky drinking days and percent change in the probability of abstinence attributed to the effect of A-CHESS operating through postdischarge service utilization. We also calculated the ratio of the mediated effect of A-CHESS (through postdischarge addiction services) to the total effect of A-CHESS regardless of mediation.
Participants were 80.2 percent white, 39.3 percent female, and had a mean age of 38.3 years. About 50.5 percent of participants reported past-month outpatient addiction treatment at any follow-up (36.0 percent at month four, 23.9 percent at month eight, and 19.9 percent at month twelve) and 75.5 percent reported past-week mutual help attendance at any follow-up (60.1 percent at month four, 60.1 percent at month eight, and 57.7 percent at month twelve).
Main Effects Analysis
When averaging the results across all time (months four, eight, and twelve), participants who received A-CHESS had significantly higher odds of outpatient addiction treatment than controls (odds ratio [OR]=2.14, 95 percent confidence interval [CI]=1.27-3.61). When looking at each follow-up period separately, rates of outpatient addiction treatment at each follow-up were approximately 9 to 11 percentage points higher in the A-CHESS arm than in the control arm (e.g., 40.4 percent versus 31.6 percent at month four).
However, when averaging the results across all follow-up periods, the odds of receiving mutual help were not significantly higher in the A-CHESS group than in controls. Looking within follow-up periods separately, A-CHESS participants received more mutual help than controls only at month twelve (not at months four and eight). Rates of mutual help at month twelve were approximately 13.6 percentage points higher in the A-CHESS arm than in the control arm.
Participants who received the A-CHESS intervention had reduced risking drinking days and increased abstinence over time compared to participants in the control group. This has been reported previously (Gustafson et al., 2014). Interestingly, the majority of outpatient addiction treatment and mutual help received at months eight and twelve was among participants who had already received those services at month four. In other words, most participants who received aftercare over the course of the whole year of follow-up had already initiated it within the first four months.
Postdischarge outpatient addiction treatment mediated the association of study arm with risky drinking days, as indicated by a statistically significant indirect effect. One way to contextualize this finding is to compare how much A-CHESS independently decreased risky drinking days to how much A-CHESS decreased risky drinking days specifically because it increased the use of outpatient addiction treatment. The effect of A-CHESS itself decreased risky drinking days by 45 percent. The ability of A-CHESS to increase postdischarge outpatient addiction treatment had an apparent effect of decreasing risky drinking days by 11 percent. Thus, the ratio of the mediated effect of A-CHESS through postdischarge outpatient addiction treatment (11 percent) to the total effect of A-CHESS (11 percent + 45 percent = 56 percent, including the direct and indirect effect) was just under 20 percent.
For the outcome of abstinence, mediation did not occur through outpatient addiction treatment or mutual help. Interestingly, A-CHESS itself was associated with both increased abstinence and increased outpatient addiction treatment, but outpatient addiction treatment was not associated with increased abstinence. However, mutual-help attendance was associated with increased abstinence.
The primary purpose of this study was to investigate the use of postdischarge addiction services as a potential mechanism of behavior change in A-CHESS, an efficacious mHealth intervention for alcohol use disorder for patients leaving residential treatment. In support of our hypothesis, A-CHESS increased the odds of outpatient addiction treatment, and the use of these treatment services was associated with reduced risky drinking days. Mediation analyses indicated that the use of postdischarge outpatient addiction treatment mediated some of the effect that A-CHESS had on risky drinking days. Specifically, A-CHESS was responsible for an 11 percent decrease in the expected number of risky drinking days across follow-ups, mediated through outpatient addiction treatment. This reduction was approximately one-fifth of the total effect of A-CHESS on risky drinking days. In contrast, mutual help did not mediate the effects of A-CHESS on risky drinking days, and the effect of A-CHESS on mutual-help services was present only at month twelve. While A-CHESS increased abstinence, neither outpatient treatment nor mutual help mediated the effects of the A-CHESS intervention on abstinence.
Our mediation analysis, which is an important first step in identifying possible mechanisms through which interventions may exert their effects (Kazdin, 2007), showed that A-CHESS was efficacious in reducing risky drinking independent of participants’ outpatient treatment utilization. Analysis also showed that A-CHESS promoted the use of outpatient treatment following discharge from residential care, which may have further reduced risky drinking.
Our study is unique in quantifying the extent to which an mHealth intervention may produce changes in individuals’ interactions with the environment—in particular, interactions with treatment systems—which may in turn lead to improved outcomes. Most of those who received addiction treatment during the latter follow-up periods had already received treatment by month four, and those who received the A-CHESS intervention were more likely to receive treatment in those late follow-up periods. This result could indicate that A-CHESS facilitates sustained engagement in aftercare. The A-CHESS trial had a relatively long intervention period (participants used A-CHESS for eight months), which could have helped sustain participants’ use of additional treatment. Aftercare is an important part of the addiction treatment continuum that may lead to improved addiction-related outcomes (Blodgett, Maisel, Fuh, Wilbourne, & Finney, 2014; McKay, 2005, 2009). We note that although approximately 20 percent of the total effect of A-CHESS on risky drinking days appeared to be through outpatient treatment, A-CHESS was not specifically designed to facilitate the use of outpatient treatment. Thus, mHealth intervention researchers and clinicians may wish to evaluate ways to facilitate ongoing care intentionally, perhaps by helping patients navigate treatment choices and overcome barriers to treatment.
There are a number of explanations for how an mHealth recovery-support intervention like A-CHESS could increase the use of outpatient treatment when provided to individuals leaving residential treatment. A-CHESS components were designed to be available anytime and anywhere, increasing the chance that participants will obtain encouragement and support to enter treatment during critical moments when they desire this type of assistance. For instance, the “Discussions” component of A-CHESS allows users to post and respond to electronic messages as a way to provide and obtain social support. This support could increase recovery-promoting behaviors, including participation in aftercare or entry into a new episode of care. Pressing the “Panic Button” feature connects users to friends, family, or other sources of support who may advocate for more treatment. The A-CHESS coach encourages participants to reach out to others when in need of support. Several other A-CHESS components (e.g., “Recovery Info,” “Our Stories”) provide information resources related to the benefits of treatment, which could also promote interest in aftercare.
Many questions remain that would help advance mHealth-supported recovery research. What constitutes meaningful use of a component: measuring all “hits” or uses of a component, or identifying specific content within each component that matters most to users? What best indicates effectiveness: measuring the number of days, times, or minutes that participants used a component? We are currently executing projects to explore such questions.
Another specific finding in our study warrants further consideration by clinicians and researchers. We observed a higher prevalence for mutual help than outpatient treatment. This result suggested that some residential treatment programs may be emphasizing mutual-help program attendance for aftercare more than formal treatment. Our results supported encouraging individuals discharging from formal treatment to attend mutual-help programs to support abstinence and long-term recovery. In addition, based on our results, we recommend suggesting both mutual help and outpatient treatment attendance to individuals discharging from higher levels of care, because our prior analyses of nationally-representative data in the United States show that participants tend to have better outcomes then they receive both mutual help and formal addiction treatment (Mowbray, Glass, & Grinnell-Davis, 2015).
Even in carefully designed randomized controlled trials, establishing causal relationships is difficult. Thus, our interpretations should not be considered as indicating that A-CHESS caused patient effects. For instance, we randomly assigned participants to receive A-CHESS support, not to receive outpatient addiction treatment or mutual help. Participants chose their own postdischarge options and the factors that influenced their choices could have influenced any reductions in drinking. Thus, we cannot definitively state that the A-CHESS mHealth application caused changes in drinking because it also caused their service use. Our measurements had limitations. To reduce participant burden, we favored a short, study-specific survey over very lengthy measures that have been subjected to more scrutiny. We also did not assess a broad range of services such as inpatient treatment, which would have been useful to consider. All measures relied upon self-report; the trial did not obtain objective measures of the outcomes (e.g., breathalyzers, medical records), though we note that several studies found sufficient matches between self-report and objective measures (Babor, Brown, & del Boca, 1990; Glass & Bucholz, 2011). In addition, while we asked counselors at the residential treatment agencies to provide care for all participants as they normally would, the counselors knew if participants were using A-CHESS or not. They might have given more attention to A-CHESS participants than controls, which might have led to improved outcomes.
Knowing how mHealth interventions influence behavior change is a critical step towards understanding how to best leverage these new technologies. The recent increase in availability of mHealth applications makes identifying effective principles of action especially important for researchers and medical health professionals who are developing mHealth resources and working with patients who use them. Future clinical trials should investigate mHealth intervention components that facilitate linking patients to needed treatment services and promote the sustained use of these services.
Babor, T. F., Brown, J., & del Boca, F. K. (1990). Validity of self-reports in applied research on addictive behaviors: Fact or fiction? Behavioral Assessment, 12(1), 5–31.
Blodgett, J. C., Maisel, N. C., Fuh, I. L., Wilbourne, P. L., & Finney, J. W. (2014). How effective is continuing care for substance use disorders? A meta-analytic review. Journal of Substance Abuse Treatment, 46(2), 87–97.
Cacciola, J. S., Alterman, A. I., Dephilippis, D., Drapkin, M. L., Valadez, C., Jr, Fala, N. C., . . . McKay, J. R. (2013). Development and initial evaluation of the Brief Addiction Monitor (BAM). Journal of Substance Abuse Treatment, 44(3), 256–263.
Glass, J. E., & Bucholz, K. K. (2011). Concordance between self-reports and archival records of physician visits: A case-control study comparing individuals with and without alcohol use disorders in the community. Drug and Alcohol Dependence, 116(1–3), 57–63.
Glass, J. E., Grant, J. D., Yoon, H. Y., & Bucholz, K. K. (2015). Alcohol problem recognition and help seeking in adolescents and young adults at varying genetic and environmental risk. Drug and Alcohol Dependence, 153, 250–7.
Glass, J. E., Perron, B. E., Ilgen, M. A., Chermack, S. T., Ratliff, S., & Zivin, K. (2010). Prevalence and correlates of specialty substance use disorder treatment for Department of Veterans Affairs Healthcare System patients with high alcohol consumption. Drug and Alcohol Dependence, 112(1–2), 150–5.
Gustafson, D. H., McTavish, F. M., Chih, M. Y., Atwood, A. K., Johnson, R. A., Boyle, M. G., . . . Shah, D. (2014). A smartphone application to support recovery from alcoholism: A randomized clinical trial. JAMA Psychiatry, 71(5), 566–72.
Kazdin, A. E. (2007). Mediators and mechanisms of change in psychotherapy research. Annual Review of Clinical Psychology, 3, 1–27.
MacKinnon, D. P. (2008). Introduction to statistical mediation analysis. New York, NY: Routledge.
McKay, J. R. (2005). Is there a case for extended interventions for alcohol and drug use disorders? Addiction, 100(11), 1594–610.
McKay, J. R. (2009). Continuing care research: what we have learned and where we are going. Journal of Substance Abuse Treatment, 36(2), 131–45.
McLellan, A. T., Lewis, D. C., O’Brien, C. P., & Kleber, H. D. (2000). Drug dependence, a chronic medical illness: Implications for treatment, insurance, and outcomes evaluation. JAMA, 284(13), 1689–95.
McTavish, F. M., Chih, M. Y., Shah, D., & Gustafson, D. H. (2012). How patients recovering from alcoholism use a smartphone intervention. Journal of Dual Diagnosis, 8(4), 294–304.
Mohr, D. C., Burns, M. N., Schueller, S. M., Clarke, G., & Klinkman, M. (2013). Behavioral intervention technologies: Evidence review and recommendations for future research in mental health. General Hospital Psychiatry, 35(4), 332–8.
Mowbray, O., Glass, J. E., & Grinnell-Davis, C. L. (2015). Latent class analysis of alcohol treatment utilization patterns and three-year alcohol-related outcomes. Journal of Substance Abuse Treatment, 54, 21–8.
Nelson, K. G., Young, K., & Chapman, H. (2014). Examining the performance of the brief addiction monitor. Journal of Substance Abuse Treatment, 46(4), 472–81.
Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. The American Psychologist, 55(1), 68–78.
Editor’s Note: This article was adapted from an article by the same authors previously published in the Journal of Substance Abuse Treatment (JSAT). This article has been adapted as part of Counselor’s memorandum of agreement with JSAT. The following citation provides the original source of the article:
Glass, J. E., McKay, J. R., Gustafson, D. H., Kornfield, R., Rathouz, P. J., McTavish, F. M., . . . Shah, D. (2017). Treatment seeking as a mechanism of change in a randomized controlled trial of a mobile health intervention to support recovery from alcohol use disorders. Journal of Substance Abuse Treatment, 77, 57–66.