Randomized controlled pilot trial of supportive text messaging for alcohol use disorder patients

Journal of Substance Abuse Treatment 94 (2018) 74–80

Contents lists available at ScienceDirect

Journal of Substance Abuse Treatment journal homepage: www.elsevier.com/locate/jsat

Randomized controlled pilot trial of supportive text messaging for alcohol use disorder patients

T

⁎

Vincent I.O. Agyaponga,b, , Michal Juhása, Kelly Mrklasc, Marianne Hraboka,d, Joy Omejeb, Irene Gladuee, Jody Kozakd, Maureen Lesliee, Pierre Chuea, Andrew J. Greenshawa a

Department of Psychiatry, University of Alberta, Edmonton, Alberta, Canada Department of Public Health, Alberta Health Services, Fort McMurray, Alberta, Canada c System Innovation and Programs, Strategic Clinical Networks™, Alberta Health Services, Alberta, Canada d Addiction and Mental Health, Alberta Health Services, Edmonton, Alberta, Canada e Northern Addiction Treatment Centre, Grande Prairie, Alberta, Canada b

A R T I C LE I N FO

A B S T R A C T

Keywords: Alcohol Technology Mental health Text message Randomized controlled trial

Aims: To evaluate the effectiveness of an addiction-related supportive text messaging mobile intervention to improve treatment outcomes for patients with alcohol use disorder (AUD). Methods: A single-rater-blinded randomized trial was conducted involving 59 AUD patients who completed a residential addiction treatment program. Patients in the intervention group (n = 29) received supportive text messages for three months following discharge. Patients in the control group (n = 30) received a text message thanking them for participating in the study. The primary outcome of this study was the three months Cumulative Abstinence Duration (CAD); secondary outcomes (units of alcohol per drinking day, numbers of days to first drink) and exploratory outcomes (health utilization) were evaluated. Subgroup analyses were also done. The enrollment rate in the study was 84%, and of those who enrolled, 73% were retained. Results: When primary and secondary outcome measures were examined via effect size analysis, the number of days to first drink was longer in the intervention than control group (large effect size, although not statistically significant). The intervention group's mean first day to drink was over twice the length of the control group (e.g., approximately 60 vs. 26 days, respectively, with a mean difference of 34.97 and 95% CI of −5.87–75.81). Small to moderate effects were found for CAD and units of alcohol per drinking day. Small to negligible effects were found for health utilization. On subgroup analyses, the participants who received text messages, among those who did not attend follow-up outpatient counselling, showed a longer CAD. Conclusions: The results suggest text messaging is a feasible and effective opportunity for follow-up care in patients discharged from residential AUD treatment.

1. Introduction Alcohol use disorder (AUD) is characterized by its chronic, recurrent nature and associated behavioural, cognitive, physiological, and social problems (American Psychiatric Association & American Psychiatric Association, 2013). In 2015, the estimated prevalence (in the past 30 days) among the adult population was 18.4% for heavy episodic alcohol use and the age-standardized prevalence of alcohol dependence was 843.2 per 100,000 people (Peacock et al., 2018). AUD is one of the most damaging, costly, and common diseases globally. According to estimates provided by the World Health Organization (WHO | Global Status Report on Alcohol and Health 2014), alcohol abuse accounts for

nearly 6% of all deaths and 5% of the global burden of disease, and is implicated in over 200 adverse health conditions. Recent Canadian alcohol consumption was estimated at 10 l per year of pure alcohol per capita for people > 15 years of age (WHO Global Information System on Alcohol and Health 2017). The total cost of AUD is staggering, particularly when considering its multiple downstream societal costs (Nutt, King, & Phillips, 2010). In Canada alone, the annual estimated cost of alcohol abuse was $14.6 billion in 2002 (Publications | Canadian Centre on Substance Use and Addiction n.d.), with AUD prevalence rates in Canada estimated at nearly 3% in 2003 (Tjepkema, 2004). A more recent population level study put the prevalence rate for problematic alcohol and/or illicit drug use in Canada at 4.6% (Dumais et al.,

⁎ Corresponding author at: Faculty of Health Sciences, Department of Psychiatry, University of Alberta, 1E1 Walter Mackenzie Health Sciences Centre (WMC), 8440 112 St NW, Edmonton, AB T6G 2B7, Canada. E-mail address: [email protected] (V.I.O. Agyapong).

https://doi.org/10.1016/j.jsat.2018.08.014 Received 17 April 2018; Received in revised form 9 August 2018; Accepted 31 August 2018 0740-5472/ © 2018 Elsevier Inc. All rights reserved.

Journal of Substance Abuse Treatment 94 (2018) 74–80

V.I.O. Agyapong et al.

an AUD-focused, supportive text messaging mobile health intervention in supporting the continuing care and recovery of patients with AUD discharged from residential treatment. Whilst there are several studies using text messages in AUD patients, this pilot study is the first of its kind in the North American context to utilize twice daily automated supportive text messages for three months in patients with AUD and incorporates a range of outcome measures at three-month follow-up. These include Cumulative Abstinence Duration (CAD; primary outcome), mean number of days to first drink and mean units of alcohol per drinking day (secondary outcomes), and health service utilization (e.g., number of visits to family physicians, psychiatrists, and other specialist physicians, Emergency Department visits, addiction counselling sessions attended; exploratory outcomes). Given the focus on identification of factors that predict participation in continuous care following residential treatment for substance use in the last decade (Arbour, Hambley, & Ho, 2011), subgroup analyses was also done to evaluate whether specific demographic and clinical factors are associated with longer CAD following use of text messages. We hypothesized that patients who received twice daily supportive text messages would have a higher Cumulative Abstinence Duration, higher days to first drink and lower units of alcohol per drinking day compared to the control group. We also hypothesized that patients receiving twice daily supportive text messages would achieve higher mean scores in respect of attendance at scheduled clinical appointments (such as with family doctors, specialists and counsellors) and achieve lower mean scores with respect to the number of emergency health services and acute care services utilized compared to patients in the control group.

2013). In addition to economic burden, AUD leads to chronic human suffering. Patients face high relapse rates and limited treatment options (Moos & Moos, 2006; Walitzer & Dearing, 2006). Additionally, patients with AUD have high psychiatric co-morbidity, including mood and anxiety disorders and are more likely to utilize acute and emergency health services (Dumais et al., 2013). Furthermore, evidence suggests AUD implicates multiple generations, with a reported 40% to 60% heritability (Agrawal et al., 2008). In light of the considerable societal burden of AUD and its recognition as a major public health problem that is associated with high relapse rates (Miller, Walters, & Bennett, 2001; Moos & Moos, 2006), innovative care options, such as use of novel technologies, is required. This is particularly true in remote and rural areas, such as Northern Alberta, where this study was conducted. In a literature review of eight studies on mobile technology-based interventions among adult users of alcohol undertaken to determine the efficacy of such interventions, the majority of studies found positive intervention effects, even though the interventions themselves varied in design, length, dosage, and target population, and were pilot or preliminary in nature (Fowler, Holt, & Joshi, 2016). Text messaging technology has been successfully implemented in reducing substance abuse (Tofighi, Nicholson, McNeely, Muench, & Lee, 2017), including programs aimed at cessation of tobacco smoking (Keoleian, Polcin, & Galloway, 2015) and with collegeage students as a means of reducing risky drinking (Rourke, Humphris, & Baldacchino, 2016). Maintenance of therapeutic contact following an intensive addiction treatment program through both traditional or mobile health technology appears to promote recovery and improve long-term clinical outcomes (Gustafson, McTavish, Chih, et al., 2014; McKay, 2005). In this respect, text messaging technology offers considerable potential as a way to provide continuous care for patients discharged from inpatient AUD treatment programs that may not otherwise be feasible due to resource and accessibility constraints. Supportive text messaging is a relatively low cost, high impact, and easily scalable program that uses existing technology, is devoid of geographic barriers, and is free and accessible to end users (Agyapong, Farren, & McLoughlin, 2011). Preliminary research using an interactive text-based help line with patients discharged from an inpatient substance treatment program suggested adequate rates of patient adherence and a statistical trend towards reduced consumption of alcohol compared to a treatment as usual group (Lucht et al., 2014). Likewise, patients with comorbid AUD and major depressive disorder who completed an inpatient treatment program responded positively to a supportive text messaging intervention (Agyapong, Ahern, McLoughlin, & Farren, 2012). In addition to symptom improvement, high patient satisfaction rates in patients with comorbid AUD and depression have been reported, with 75% of patients reporting text messages reminded them to abstain from alcohol, and 83% endorsing the messages as motivating recovery and preventing relapse (Agyapong, Milnes, McLoughlin, & Farren, 2013). An exploratory, single-blind randomized controlled pilot study comparing four different types of alcohol reduction-themed text messages sent daily to weekly drink self-tracking texts sought to determine their impact on drinking outcomes over a 12-week period. The study found that almost 80% of study participants wanted to continue receiving messages for an additional 12 weeks at the end of the study (Muench et al., 2017), suggesting high acceptability and feasibility of text messaging interventions for AUD. Other positive outcomes related to medication adherence, engagement with peer support groups appointment attendance, motivation, self-efficacy, relapse prevention and social support were documented in a recent systematic review of mobile phone messaging interventions for illicit drug and alcohol dependence (Tofighi et al., 2017). These findings reinforce the feasibility and acceptability of using text-message interventions to support patients with AUD. Existing studies provide preliminary support for the feasibility, acceptability, and effectiveness of text messaging following inpatient treatment for AUD. The objective of this study was to test the initial efficacy of using

2. Methods 2.1. Study design and participants This study was a single-rater blinded randomized trial of daily addiction-related supportive text messages delivered to participants' mobile phones. Participants were recruited from patients completing the 28-day addiction treatment program at the Northern Addiction Rehabilitation Centre in Grande Prairie, Alberta, Canada, from June 2015 to December 2015. The residential addiction treatment program consists of one-to-one, group, and family counselling, interactive workshops, information sessions, recreation and leisure programming, nutrition assessment and support, self-help and relapse prevention groups, discharge planning and after program support. The individual and group programs are run by addiction counsellors and social workers and patients have access to psychiatric consultation. Written and oral informed consent was obtained from each participant. The study protocol was approved by the Research Ethics Board of the University of Alberta and published (Agyapong et al., 2015). The study was conducted in accordance with the Declaration of Helsinki (World Medical Association, 2013) and WHO Good Clinical Practice Guidelines (Guidelines for Good Clinical Practice (GCP) for Trials on Pharmaceutical Products. WHO Technical Report Series, No. 850, Annex 3 - WHO Expert Committee on Selection and Use of Essential Medicines, Sixth Report, 1993 n.d.). The trial was registered with clinicaltrials.gov (NCT02327858). CONSORT criteria were used for reporting study findings (Schulz et al., 2010). Study participants met the following inclusion criteria: 1. Age 18 years and above and capable of providing informed consent. 2. Completing the final week of admission at the 28-day residential addiction treatment program at the Northern Addiction Treatment Centre in Grande Prairie, Alberta, and fulfilled the DSM-5 diagnostic criteria for Alcohol Use Disorder (American Psychiatric Association & American Psychiatric Association, 2013). Diagnosis was determined following a structured clinical interview by a licensed addiction counsellor to elicit the presence or absence of specific diagnostic criteria, as outlined in the DSM-5. 75

Journal of Substance Abuse Treatment 94 (2018) 74–80

V.I.O. Agyapong et al.

Fig. 1. Study flow diagram.

After baseline assessment, participants were randomized using a series of random numbers generated using Microsoft Excel. Participants were allocated the next available number from the random number sequence by a research associate, who did not participate in follow-up assessments or analysis. Participants were placed in the intervention or control group respectively, depending on whether the randomization number was even or odd. The research assistant who performed or supervised the follow-up assessments remained blind to group allocation throughout the study period.

3. Patients had a text-message enabled mobile phone, were familiar with text messaging technology (could retrieve text messages), were able to read English, and were available for follow-up during the study period. Patients who did not have mobile phones or refused to consent were excluded from the study. 2.2. CONSORT 2010 flow diagram Seventy patients were assessed for eligibility to enter the trial, of which 59 eligible patients were randomized (Fig. 1). As summarized in Fig. 1, 29 participants were allocated to the intervention group and 30 to the control group. Of the enrolled participants, 21 and 22 patients in the intervention and control groups, respectively, completed the study. Of those who withdrew consent for follow-up, no specific explanation was provided for their decision. Of note, the enrollment rate in the study was 84%, and of those who enrolled, 73% were retained for the duration of the study. The two patients who were not eligible did not fulfil the DSM 5 criteria for AUD.

2.4. Interventions Starting one day after discharge from the residential addiction treatment facility, patients in the intervention group received twice daily, automated, AUD-related supportive text messages for a threemonth period. Text messages were formulated based on addiction treatment principles to target avoidance of alcohol use. These were composed by addiction counsellors in collaboration with patients, who met the inclusion criteria but were not a part of the current study. Relevant text messages were solicited from addiction counsellors and patients receiving addiction and mental health services in Fort McMurray through an open poster invitation. The messages received were reviewed and refined into a maximum of 160-word characters by an expert group comprising a psychiatrist, an addiction counsellor and a social worker. The messages were then pre-programed into online software that delivered text messages at 10.00 h and 19.00 h (Mountain Time) each day. The time 10.00 h and 19.00 h were selected for the delivery of the text messages due to patients expressing satisfaction

2.3. Baseline procedures, randomization and masking Patients in their final week of admission to the 28-day residential addiction treatment program in Grande Prairie were approached by an addiction counsellor and invited to consider participating in the trial. Eligible consenting patients completed baseline assessments, which included collection of demographic and clinical information. 76

Journal of Substance Abuse Treatment 94 (2018) 74–80

V.I.O. Agyapong et al.

their assessments with a research assistant over the phone. The research assistant read the assessment questions aloud and asked participants to indicate her/his responses. Our primary research hypothesis was that the provision of twicedaily supportive text messages would significantly increase the threemonth Cumulative Abstinence Duration (CAD) since completion of the residential treatment facility (baseline) compared to the control group. The primary outcome measure was the CAD at three months from baseline and was assessed using the timeline follow-back procedures described by Sobell and Sobell (Litten & Allen, 1992). Secondary outcome measures included mean number of days to first drink and mean units of alcohol per drinking day, assessed by asking patients about the types and quantities of alcohol they consumed for each of the days they recalled drinking. Exploratory outcome measures included self-reports of the number of times participants utilized health services, including the number of visits to family physicians, psychiatrists, and other specialist physicians. It also included self-reports of the number of visits to the Emergency Department, the number of addiction counselling sessions attended, and the number of other health services utilized.

Table 1 Distribution of baseline demographic and clinical characteristics of participants. Variable

Intervention group N

Control group N

p-Value

Gender

22 (75.9%) 7 (24.1%) 16 (57.1%) 12 (42.9%) 39.4 (SD = 10.6) 11 (37.9%)

22 (73.3%) 8 (26.7%) 15 (51.7%) 14 (48.3%) 41.8 (SD = 10.0) 13 (43.3%)

1.0

18 (62.1%)

17 (56.7%)

21 (72.4%) 8 (27.6%)

23 (76.7%) 7 (23.3%)

0.77

14 (48.1%)

18 (60.0%)

0.60

13 (51.9%)

12 (40.0%)

11 (42.3% 15 (57.7%) 7 (25.0%) 21 (75.0%)

13 (46.4%) 15 (53.6%) 8 (27.6%) 21 (72.4%)

0.79

23.0 (SD = 7.5) 21.8 (SD = 55.8)

25.0 (SD = 10.8) 23.7 (SD = 55.7)

0.47

13 (50.0%) 13 (50.0%)

9 (31.0%) 20 (69.0%)

0.18

Male Female Age ≤40 ≥41 Mean age in years (SD = standard deviation) Formal educational Up to high level school College/ university Employment status Employed Not employed Relationship status In a relationship Not in a relationship On antidepressants Yes before enrolment No Yes On medication for No chronic physical health problems Mean age of onset of problem drinking in years Mean longest duration of sobriety since onset of problem drinking in months Past treatment for Yes alcohol abuse No

0.79 0.38 0.79

2.6. Statistical analysis

1.0

In this pilot trial, we utilized data elicited from participants who could reasonably be enrolled within the existing budget and time frame. Trial enrolment reached a total of 59 participants, which closely approximated the initial target sample size of 60 reported in our published alcohol trial protocol (Agyapong et al., 2015). Data were analysed on an intention-to-treat basis using SPSS version 20 for Windows (IBM SPSS - United States, n.d.). Baseline demographic and clinical characteristics of intervention and control groups were analysed using Chi-squared, Fisher's Exact, and Student's t-tests. Threemonth CAD was compared between the intervention and control groups using the Student's t-test. Participants with missing follow-up data were balanced between the two groups (8 in the text message group and 8 in the control group). Consequently, we imputed CAD values of zero, which assumes the participant relapsed on the first day and drank daily for three months (Molenberghs & Kenward, 2008). To explore the impact of select demographic and clinical variables on the three-month CAD, we performed a two-way between group analysis of variance. Secondary and exploratory outcome measures were compared using Student's t-tests and effect size analyses.

0.49

with this timing in a previous study (Agyapong et al., 2013). The online program allowed verification that messages were delivered to patients but could not confirm messages had been read. Two different messages were sent each day; no messages were repeated within the 90-day period. The same messages were delivered to each patient, determined solely according to the day they joined the text support program. Patients were informed as part of the first message not to reply to any of the messages, as responding could incur a charge. Examples of text messages included the following:

• Think of your recovery as an opportunity to find new solutions in your • • •

3. Results

life. Remember that the past is gone and what you do next is what really matters. Before you think of the next drink, think of the last one and how it made you feel. The thoughts of drinking will return to test you. Remember to never give up in the face of temptation. The true joy of recovery is in the fellowship and camaraderie you share with others travelling the same path.

3.1. Baseline demographic and clinical characteristics As shown in Table 1, baseline demographic and clinical characteristics were similar in both treatment groups. Both groups were predominantly male, had an average age of approximately 40, attained a high school or post-secondary education, and were employed. A large proportion had previously used antidepressant medication, and approximately a quarter of participants were taking medication for a chronic medical condition. In terms of drinking history, participants had been drinking in a problematic manner for 15 years on average, and many had sought previous treatment for alcohol use.

Patients in the control group received the same text messages once each fortnight reading “Thank you for participating in our study”. All study participants, regardless of study condition, were encouraged to participate in the usual follow-up care associated with completing a residential addiction treatment program including attending outpatient addiction counselling and visiting their regular doctors or the emergency room, as needed.

3.2. Primary and secondary outcome measures Table 2 indicates that primary and secondary outcomes measures did not reach statistical significance threshold (p < 0.05) for differences between groups, although a notable statistical trend persisted with the intervention group, maintaining abstinence for a much higher number of days (CAD) after discharge from the residential treatment program. When secondary outcomes were examined in terms of effect sizes, the number of days to first drink was longer in the intervention than control group, with a large effect noted according to Cohen's

2.5. Procedures at follow-up, outcome measures, and hypothesis At three months following their enrolment, all participants were contacted by a blinded research assistant who assisted them in completing assessment questions related to the primary and secondary outcome measures listed below. All the study participants completed 77

Journal of Substance Abuse Treatment 94 (2018) 74–80

V.I.O. Agyapong et al.

Table 2 Mean scores and group difference statistics for the primary and secondary outcome measures for the intervention and control groups at three-month follow-up. Variable

Intervention group

Control group

t

p-Value

Mean difference

Confidence intervals

Cohen's da

Cumulative abstinence duration (CAD) at three months in days Number of days to first drink Units of alcohol per drinking days

83.5 (SD = 19.3) 60.8 (SD = 31.2) 1.0 (SD = 2.3)

73.6 (SD = 33.0) 25.8 (SD = 28.7) 1.5 (SD = 2.8)

1.16 1.94 −0.60

0.25 0.09 0.55

9.88 34.97 −0.48

−7.32–27.08 −5.87–75.81 −2.11–1.14

0.37 1.17 −0.20

a

Cohen's d computed as Cohen's d = (M2 − M1) / SDpooled, where SD pooled is √((SD12 + SD22) / 2) (Effect Size Calculator (Cohen's D) for t-Test n.d.).

Table 3 Health services utilization at three months for the intervention and control groups. Variable Mean Mean Mean Mean Mean a

number number number number number

of of of of of

visits to family physician visits to psychiatrist visits to other specialists visits to the Emergency Department counselling sessions attended

Intervention group

Control group

t

p-Value

Mean difference

95% confidence intervals

Cohen's da

1.3 (SD = 2.1) 1.0 (SD = 2.7) 0.24 (SD = 0.62) 0.24 (SD = 0.54) 3.0 (SD = 3.3)

1.7 (SD = 1.9) 0.64 (SD = 1.2) 0.45 (SD = 2.1) 0.41 (SD = 1.1) 4.7 (SD = 5.6)

−0.65 0.48 −0.45 −0.64 −1.18

0.52 0.63 0.66 0.52 0.25

−0.40 0.36 −0.22 −0.17 −1.68

0.61–(−1.63) 0.75–(−1.15) −1.12–0.76 −0.71–0.36 −4.57–1.21

−0.20 0.17 −0.14 −0.20 −0.37

Cohen's d computed as Cohen's d = (M2 − M1) / SDpooled, where SD pooled is √((SD12 + SD22) / 2) (Effect Size Calculator (Cohen's D) for t-Test n.d.).

achieved a higher mean CAD when they received AUD-related supportive text messages compared to control participants [i.e. 88 (SD = 4.6) versus 53.4 (SD = 43.5), respectively].

conventions (Cohen, 1977). On qualitative inspection of the data, the intervention group's mean first day to drink was over twice the length of the control group (e.g., approximately 60 vs. 26 days, respectively). A small to moderate effect of longer CAD at three-month follow-up was noted for the intervention group, and a small effect to consume less alcohol per drinking day was noted for the intervention group. The results of self-reported health utilization are depicted in Table 3. Differences in health utilization were not significant, with overall small to negligible effect sizes according to Cohen's conventions (Cohen, 1977). An exception was the small to moderate effect size for the mean number of counselling sessions attended, with the control group attending more sessions than the intervention group. When the groups were examined subdivided based on demographic or clinical variables (see Table 4) subgrouping based on attendance at outpatient addiction counselling demonstrated a significant interaction effect. Specifically, participants who had no outpatient counselling

4. Discussion This pilot RCT evaluated the initial efficacy of a supportive text messaging intervention to promote alcohol avoidance in recovering AUD patients after discharge from a residential treatment program into the community. Recovering AUD patients returning to their everyday lives face many challenges, stressors, and alcohol-related cues that can increase alcohol cravings and - in many cases - can lead to a relapse (Miller et al., 2001; Moos & Moos, 2006). Long-term clinical contact to promote continued abstinence and community integration can be costly and difficult to access - especially in remote and rural areas, such as Northern Alberta.

Table 4 Two-way ANOVA comparison between patient characteristics and supportive text messaging intervention on Cumulative Abstinence Duration (CAD). Independent variables

Gender Age Formal educational level Employment status Relationship status On chronic physical health medication On antidepressants Previous treatment for alcohol abuse Outpatient addiction counselling sessions I

Outpatient addiction counselling sessions II

Male Female ≤40 ≥41 Up to high school College/university Employed Not employed In a relationship Not in a relationship Yes No Yes No Yes No Had no CBT/ counselling Had at least one CBT/ counselling session Had no addiction counselling Had one to three addiction counselling sessions Had four or more addiction counselling sessions

Intervention group mean CAD (SD)

Control group mean CAD (SD)

Treatment interaction F

Treatment interaction p

Subgroup main effect F

Subgroup main effect p

87.7 (SD = 7.7) 70.4 (SD = 35.9) 79.1 (SD = 27.4) 89.7 (SD = 0.9) 89.2 (SD = 1.3) 81 (SD = 22.9) 87.0 (SD = 8.2) 72.8 (SD = 36.8) 88.6 (SD = 3.9) 79.25 (SD = 29.2) 89.6 (SD = 0.82) 80.07 (SD = 23.6) 77.8 (SD = 27.8) 89.7 (SD = 1.0) 76.0 (SD = 30.8) 90.0 (SD = 0.0) 88 (SD = 4.6)

75.9 (32.2) 69 (SD = 36.7) 81.0 (SD = 28.5) 64.5 (SD = 37.8) 63.4 (SD = 42.8) 82.8 (SD = 18.2) 76.8 (SD = 31.3) 65.5 (SD = 38.9) 78.1 (SD = 28.0) 64.4 (SD = 43.2) 58.0 (SD = 43.8) 79.0 (SD = 27.8) 73.2 (SD = 31.6) 73.9 (SD = 33.8) 64.7 (SD = 43.2) 77.1 (SD = 28.3) 53.4 (SD = 43.5)

0.31

0.6

1.6

0.2

2.4

0.13

0.11

0.74

2.5

0.13

0.4

0.5

0.02

0.88

1.74

0.2

0.06

0.80

1.65

0.21

2.6

0.12

0.36

0.55

0.53

0.47

0.65

0.43

0.01

0.93

1.87

0.18

7.56

0.01

2.84

0.10

79.5 (SD = 25.7)

88.7 (SD = 4.6)

88 (SD = 4.6)

53.4 (SD = 43.5)

4.08

0.03

1.42

0.26

89.5 (SD = 1.0)

90.0 (SD = 0.0)

73.9 (SD = 39.5)

88.4 (SD = 5.1)

78

Journal of Substance Abuse Treatment 94 (2018) 74–80

V.I.O. Agyapong et al.

Our study reported similar findings as previous preliminary research showing that supportive text messaging can be an effective tool to support patients' avoidance of alcohol use on discharge from a residential AUD program (Agyapong et al., 2012, 2017; Lucht et al., 2014). Our study provides evidence that twice daily supportive text messaging is feasible and potentially effective in a geographically remote and rural area in a local context. Importantly, this study examines the effect of twice daily supportive text messaging on several outcomes, including alcohol use variables as well as health utilization, and the effect of demographic and clinical factors on the utility of text messaging. When secondary outcomes were examined, the group who received text messages had twice the length of time interval to first drink of alcohol on average, than the control group who did not receive text messages, with small to moderate effects found in favour of the intervention group for longer CAD and less alcohol consumption. Moreover, the enrollment and retention rate of the study (84%, 73%, respectively) may suggest high patient acceptability of the intervention, which is important in a patient population with high levels of comorbidity, variable levels of previous treatment success, and fluctuating levels of readiness for change. In this study, small to negligible effects of text messaging on health utilization was found, except for a small to moderate effect indicating people who received text messages attended fewer outpatient counselling sessions. The limited effects of text messaging on health utilization variables may be due in part to the complexity of the sample, who despite their young age, presented with a long history of problematic alcohol use and clinical comorbidity, both in terms of psychiatric (e.g., use of antidepressants) and physical (chronic health) conditions. This may have resulted in a need for higher utilization of health services overall. When the sample was sub grouped according to demographic and clinical variables, a significant interaction effect emerged. Specifically, among participants who attended no outpatient counselling after discharge from residential treatment, those who received text messages had significantly longer period of abstinence from drinking (by about 65%). One potential implication of this finding is that the participants who did not attend follow-up counselling found that the text messages were sufficient in supporting their recovery. This is a hypothesis that would require further exploration in subsequent studies. Although not a replacement for treatment, the results of this study suggest that supportive text messaging offers a viable tool to provide supportive care to patients in recovery. These people may be challenged to reintegrate into the community following an intensive and immersive residential treatment stay. Given the scope of AUD as a major public health problem, a simple tool that is highly cost-effective, easily scalable, devoid of practical barriers (e.g., geography, computer ownership), and clinically effective in improving outcomes, offers one clear and apparent solution to improving the continuity of care in people with AUD. Future research using supportive text messaging may wish to include tailored messaging based on risk factors for relapse, such as depression (Suter, Strik, & Moggi, 2011). For example, specific moodfocused text messaging content could be included for persons with comorbid AUD and depressive symptoms. Limitations of this study exist that bear mention. First, like previous research (Agyapong et al., 2012; Bock et al., 2016; Haug, Lucht, John, Meyer, & Schaub, 2015; Lucht et al., 2014), this study was underpowered in that a lack of statistical significance was noted, despite the large effect sizes associated with some outcomes (e.g., time interval to first drink). This compels the need for a larger randomized control trial that may involve multiple treatment centers across geographical regions and a longer recruitment period. Second, incorporation of a patient satisfaction measure may better help elucidate the acceptability of the intervention among patients. Third, this study was conducted in patients discharged from a residential treatment program. Thus, findings represent a specific group of patients and may not be generalizable

to persons with a shorter drinking history and lower levels of clinical complexity. Finally, our study reports only three month outcomes and therefore it is not possible to know if the intervention had the same effects at one or two months, and/or if any positive trends observed at 3 months would be sustained after the intervention ended. It is worth noting that the efforts to collect six-month outcomes data as well as data related to acceptability and patient satisfaction with the intervention were hampered by displacement of the research team by the May 2016 wild fires in Fort McMurray and it is hoped a future study would explore the early and long-term effects of the intervention as well as patient satisfaction. Acknowledgements We will like to express our sincere gratitude to addiction councilors and staff of the Northern Addiction Treatment Centre in Grande Prairie for supporting recruitment of study participants. Conflict of interest All authors declare we have no conflict of interest to declare. Funding This study was funded by a quality improvement grant from Alberta Health Services. References Agrawal, A., Hinrichs, A. L., Dunn, G., Bertelsen, S., Dick, D. M., Saccone, S. F., ... Bierut, L. J. (2008). Linkage scan for quantitative traits identifies new regions of interest for substance dependence in the Collaborative Study on the Genetics of Alcoholism (COGA) sample. Drug and Alcohol Dependence, 93(1–2), 12–20. Agyapong, V. I. O., Ahern, S., McLoughlin, D. M., & Farren, C. K. (2012). Supportive text messaging for depression and comorbid alcohol use disorder: Single-blind randomised trial. Journal of Affective Disorders, 141(2–3), 168–176. Agyapong, V. I. O., Farren, C. K., & McLoughlin, D. M. (2011). Mobile phone text message interventions in psychiatry - What are the possibilities? Current Psychiatry Reviews, 7(1), 50–56. Agyapong, V. I. O., Juhás, M., Ohinmaa, A., Omeje, J., Mrklas, K., Suen, V. Y. M., ... Greenshaw, A. J. (2017). Randomized controlled pilot trial of supportive text messages for patients with depression. BMC Psychiatry, 17(1), 286. Agyapong, V. I. O., Milnes, J., McLoughlin, D. M., & Farren, C. K. (2013). Perception of patients with alcohol use disorder and comorbid depression about the usefulness of supportive text messages. Technology and Health Care. 21(1), 31–39. Agyapong, V. I. O., Mrklas, K., Suen, V. Y. M., Rose, M. S., Jahn, M., Gladue, I., ... Greenshaw, A. (2015). Supportive text messages to reduce mood symptoms and problem drinking in patients with primary depression or alcohol use disorder: Protocol for an implementation research study. JMIR Research Protocols, 4(2), e55. American Psychiatric Association, & American Psychiatric Association (Eds.). (2013). Diagnostic and Statistical Manual of Mental Disorders: DSM-5(5th ed.). Washington, D.C.: American Psychiatric Association. Arbour, S., Hambley, J., & Ho, V. (2011). Predictors and outcome of aftercare participation of alcohol and drug users completing residential treatment. Substance Use & Misuse, 46(10), 1275–1287. Bock, B. C., Barnett, N. P., Thind, H., Rosen, R., Walaska, K., Traficante, R., ... ScottSheldon, L. A. J. (2016). A text message intervention for alcohol risk reduction among community college students: TMAP. Addictive Behaviors, 63, 107–113. Cohen, J. (1977). Statistical power analysis for the behavioral sciences (Rev. ed.). New York: Academic Press. Dumais, A., De Benedictis, L., Joyal, C., Allaire, J. F., Lesage, A., & Côté, G. (2013). Profiles and mental health correlates of alcohol and illicit drug use in the Canadian population: An exploration of the J-curve hypothesis. Canadian Journal of Psychiatry, 58(6), 344–352. https://doi.org/10.1177/070674371305800606. Effect size calculator (Cohen's d) for t-test. (2016). Retrieved from http://www. socscistatistics.com/effectsize/Default3.aspx>, Accessed date: 22 November 2017 (n.d.). Fowler, L. A., Holt, S. L., & Joshi, D. (2016). Mobile technology-based interventions for adult users of alcohol: A systematic review of the literature. Addictive Behaviors, 62, 25–34. https://doi.org/10.1016/j.addbeh.2016.06.008. Guidelines for good clinical practice (GCP) for trials on pharmaceutical products. WHO technical report series, no. 850, annex 3 - WHO Expert Committee on Selection and Use of Essential Medicines, sixth report, 1993. (2014). Retrieved from http://apps. who.int/medicinedocs/en/d/Jwhozip13e/, Accessed date: 22 November 2017 (n.d.). Gustafson, D. H., McTavish, F. M., Chih, M., et al. (2014). A smartphone application to support recovery from alcoholism: A randomized clinical trial. JAMA Psychiatry. 71(5), 566–572. https://doi.org/10.1001/jamapsychiatry.2013.4642.

79

Journal of Substance Abuse Treatment 94 (2018) 74–80

V.I.O. Agyapong et al.

(2018). Global statistics on alcohol, tobacco and illicit drug use: 2017 status report. Addiction. https://doi.org/10.1111/add.14234. Publications | Canadian Centre on Substance Use and Addiction. (2016). Retrieved from http://www.ccdus.ca/Eng/resources/Pages/default.aspx#k=all(The%20costs%20of %20substance%20abuse%20in%20Canada%202002), Accessed date: 23 November 2017 (n.d.). Rourke, L. O., Humphris, G., & Baldacchino, A. (2016). Electronic communication based interventions for hazardous young drinkers: A systematic review. Neuroscience and Biobehavioral Reviews, 68, 880–890. Schulz, K. F., Altman, D. G., Moher, D., & CONSORT Group (2010). CONSORT 2010 statement: Updated guidelines for reporting parallel group randomised trials. BMJ, 340, c332 (mar23 1). Suter, M., Strik, W., & Moggi, F. (2011). Depressive symptoms as a predictor of alcohol relapse after residential treatment programs for alcohol use disorder. Journal of Substance Abuse Treatment, 41(3), 225–232. Tjepkema, M. (2004). Alcohol and illicit drug dependence. Health Reports, 15, 9–19 (Suppl). Tofighi, B., Nicholson, J. M., McNeely, J., Muench, F., & Lee, J. D. (2017). Mobile phone messaging for illicit drug and alcohol dependence: A systematic review of the literature. Drug and Alcohol Review, 36(4), 477–491. Walitzer, K. S., & Dearing, R. L. (2006). Gender differences in alcohol and substance use relapse. Clinical Psychology Review, 26(2), 128–148. WHO Global Information System on Alcohol and Health (pp. –). (2017). Geneva: World Health Organization. Retrieved from http://apps.who.int/gho/data/node.main. GISAH?showonly=GISAH, Accessed date: 22 November 2017. WHO | Global Status Report on Alcohol and Health 2014. (2013). Retrieved from http:// www.who.int/substance_abuse/publications/global_alcohol_report/en/, Accessed date: 22 November 2017 (n.d.). World Medical Association (2013). World Medical Association Declaration of Helsinki: Ethical principles for medical research involving human subjects. JAMA, 310(20), 2191–2194.

Haug, S., Lucht, M. J., John, U., Meyer, C., & Schaub, M. P. (2015). A pilot study on the feasibility and acceptability of a text message-based aftercare treatment programme among alcohol outpatients. Alcohol and Alcoholism (Oxford, Oxfordshire), 50(2), 188–194. IBM. How to cite IBM SPSS Statistics or earlier versions of SPSS - United States. (2015). Retrieved from http://www-01.ibm.com/support/docview.wss?uid=swg21476197, Accessed date: 22 November 2017 (n.d.). Keoleian, V., Polcin, D., & Galloway, G. P. (2015). Text messaging for addiction: A review. Journal of Psychoactive Drugs, 47(2), 158–176. Litten, R. Z., & Allen, J. P. (1992). Measuring alcohol consumption: Psychosocial and biochemical methods. Totowa, NJ: Humana Presshttps://doi.org/10.1007/978-1-46120357-5 (online, 23 November 2017). Lucht, M. J., Hoffman, L., Haug, S., Meyer, C., Pussehl, D., Quellmalz, A., ... Schomerus, G. (2014). A surveillance tool using mobile phone short message service to reduce alcohol consumption among alcohol-dependent patients. Alcoholism, Clinical and Experimental Research, 38(6), 1728–1736. McKay, J. R. (2005). Is there a case for extended interventions for alcohol and drug use disorders? Addiction (Abingdon, England), 100(11), 1594–1610. Miller, W. R., Walters, S. T., & Bennett, M. E. (2001). How effective is alcoholism treatment in the United States? Journal of Studies on Alcohol, 62(2), 211–220. Molenberghs, G., & Kenward, M. G. (2008). Missing data in clinical studies. (Reprinted Statistics in practice. Chichester: Wiley (2008)). Moos, R. H., & Moos, B. S. (2006). Rates and predictors of relapse after natural and treated remission from alcohol use disorders. Addiction (Abingdon, England), 101(2), 212–222. Muench, F., van Stolk-Cooke, K., Kuerbis, A., Stadler, G., Baumel, A., Shao, S., ... Morgenstern, J. (2017). A randomized controlled pilot trial of different mobile messaging interventions for problem drinking compared to weekly drink tracking. PLoS One, 12(2), e0167900. https://doi.org/10.1371/journal.pone.0167900. Nutt, D. J., King, L. A., & Phillips, L. D. (2010). Drug harms in the UK: A multicriteria decision analysis. The Lancet, 376(9752), 1558–1565. Peacock, A., Leung, J., Larney, S., Colledge, S., Hickman, M., Rehm, J., ... Degenhardt, L.

80