Nicotine & Tobacco Research, Volume 16, Number 12 (December 2014) 1586–1592
Original Investigation
How Do Smokers Use a Smoking Cessation Text Messaging Intervention? James Balmford PhD, Ron Borland PhD VicHealth Centre for Tobacco Control, Cancer Council Victoria, Melbourne, Australia Corresponding Author: James Balmford, PhD, VicHealth Centre for Tobacco Control, Cancer Council Victoria, 615 St. Kilda Road, Melbourne, VIC 3004, Australia. Telephone: 61-3-9514-6291; E-mail: [email protected] Received March 5, 2014; accepted May 31, 2014
Abstract
Methods: QuitTxt was offered during participation in a previously reported randomized controlled trial and was activated by 924 smokers or recent quitters, of whom 862 used it to a criterion level. Outcome data (quit attempts, 7-day point prevalence abstinence, and perceived helpfulness) were collected 1 month after first use. Results: Most (68.9%) accepted the default of 4–8 messages per day, and median use duration was 27 days. Half (49.1%) appeared to miss reporting at least 1 status change, with relapses less likely to be reported than quit progression. Emergency help was used by 27.0% of those eligible for it; emergency help was used more frequently among those with recent quit experience and lower nicotine dependence. Use of emergency help was unrelated to short-term cessation outcome. Conclusions: The most notable finding is the variability in use. Some users complied fully with the requirement to report status changes, while even among those who did not, many found QuitTxt to be very helpful, suggesting that perfect congruence between message content and quit status is not essential. The use of emergency help functionality was relatively rare but was appreciated.
Mobile-phone-based interventions of text (SMS) and/or video messaging are an effective method of delivering smoking cessation assistance, with a pooled risk ratio of 1.71 (95% CI = 1.47– 1.99) across five studies (Whittaker et al., 2012) including the text messaging study of ours that is elaborated on in this paper (Borland, Balmford, & Benda, 2013). These programs provide brief snippets of advice to help prompt action, strategies to resist temptations to smoke, timely reminders to prepare for upcoming challenges, and facilitate persistence in a quit attempt. A unique benefit is their ease of use anywhere at any time, while shared benefits with other automated programs include the capacity for tailoring, appeal to young people, and the potential for high reach at low cost-per-user. While the delivered content of these programs is specified, little is known about how these programs are actually used, including to what extent interactive components are utilized. To the extent that programs are not used as intended, their effectiveness may be reduced. Understanding how programs are used is necessary if we are to shape them to be both effective and maximally acceptable to users. In most published reports of texting programs (e.g., Free et al., 2011; Rodgers et al., 2005), the program requires a quit
date and provides messages assuming that the person quits on that date and stays stopped. Some recent programs (e.g., Haug, Meyer, Schorr, Bauer, & John, 2009; Snuggs et al., 2012) send periodic active requests for status information, thereby ensuring appropriate ongoing tailoring of content, while ours (Balmford, Borland, Benda, & Howard, 2013; Borland et al., 2013) requires sending specified text messages to notify of changes in situation (setting a date, initiating a quit attempt, and relapse). What little evidence there is suggests that smokers tend to comply with requests for status information, particularly if they have progressed in their quit attempt as opposed to relapsed (Snuggs et al., 2012). The timing of these requests is unlikely to correspond with the timing of actual changes, however, and it is not known to what extent users will report changes unprompted at the time they occur. Text messaging programs can also provide “emergency help” in response to user-initiated requests for support. This is a component of most mobile-phone-based interventions for smoking cessation, and potential users want and expect such functionality (Bock, Heron, Jennings, Magee, & Morrow, 2012). However, little is known about its perceived helpfulness
doi:10.1093/ntr/ntu111 Advance Access publication July 16, 2014 © The Author 2014. Published by Oxford University Press on behalf of the Society for Research on Nicotine and Tobacco. All rights reserved. For permissions, please e-mail: [email protected].
1586
Downloaded from http://ntr.oxfordjournals.org/ at University of North Dakota on July 8, 2015
Introduction: Mobile-phone-based text (SMS) messaging is an effective method for delivering smoking cessation assistance; however, little is known about optimal program use. This paper reports on the use of 2 forms of interaction (reporting changes in quit status and emergency help) among users of QuitTxt, an interactive, automated text messaging advice program. We examined preferences for messaging intensity, duration of use, and their associations with short-term cessation outcome or perceived helpfulness.
Nicotine & Tobacco Research
Methods Participants and Procedure The sample consisted of 924 smokers or recent quitters who registered with and subsequently activated the QuitTxt program to begin receiving messages during a study of “how effective Internet- and telephone-based resources are in helping smokers quit.” This represented 40.2% of the 2,299 participants from three arms of the study randomized to an offer of QuitTxt alone, or two ways of offering it in conjunction with the QuitCoach, an Internet-based expert system that provides tailored advice (Borland, Balmford, & Hunt, 2004). Study details, including outcomes, are provided in the study by Borland and colleagues (2013). Briefly, study participants were recruited from
help seekers (typically quitline callers, n = 869) or were coldcontacted from survey panels provided by a Melbourne-based market research company (n = 1,430). Over half (63.3%) of the 924 QuitTxt registrants came from the help-seeker group. Most of the help-seeker sample completed the recruitment assessment over the phone, whereas the entire cold-contacted sample responded to an E-mail invitation from the survey company and completed it online. A number of demographic- (age and gender) and smoking-related variables (nicotine dependence as measured by number of cigarettes per day and quit attempts in the past year) were measured at baseline, as was perceived comfort using SMS messaging. Progress to quit was categorized from standard questions on smoking status as no quit date set, quit date set, and already quit. QuitTxt Description and Signup QuitTxt is designed to be used by all smokers regardless of their current progress to quit and by recent quitters (up to 1-month quit). For those without a set quit date, the focus of messages is on motivating an attempt and gaining a commitment. For those with a quit date, QuitTxt is structured to lead the person to successful implementation, with a focus of activity in the day before the attempt. Once quit the focus moves to preventing relapse, initially by presenting strategies and motivational messages, but with more of an emphasis on achieving milestones over time. Three message intensities are available: default (3–8 messages per day), reduced (2–6 messages per day), and light (1–4 messages per day). Message intensity peaks around times of changes in status, for example, when one sets a quit date, around the quit date, and following relapse (if that happens), and then is gradually reduced. Tailoring of messages is primarily to progress to quit, with some other variables (e.g., age, gender, nicotine dependence) providing additional tailoring. The program is stopped at 1-month quit, if it is apparent that the user is not progressing (20 consecutive days without having set a quit date), or by request (by texting “STOP”). In the trial, all users were mailed a wallet-sized card with the commands recognized by QuitTxt to report changes in status, and mailed and/or E-mailed an A4-sized annotated version of the card. “Emergency help” is also offered to those who are quit, via several commands (e.g., STRESS, SOCIAL, BORED), with a coping response sent immediately, typically followed half an hour later by a message designed to encourage reflection on how well the strategy worked. Participants can also customize timing of the first and last messages of the day (messages are evenly spaced between these times), and whether they want different start and end times for weekdays and weekends. Follow-Up Data The 1-month follow-up, conducted via an E-mailed web link with phone follow-up for nonresponders, established whether participants had made a quit attempt since baseline, 7-day point prevalence abstinence, and other indicators of quitting activity. The 96.6% who reported evidence of quitting activity since baseline (a quit attempt, reduced consumption, or thoughts about quitting) were asked “How helpful was the program of text messages that you used?” and answered very helpful, somewhat helpful, and not at all helpful. Reported usage and status were combined with electronic records to estimate the extent of inadequate use; that is, failure to notify the program of status changes. The definition of inadequate use differed by
1587
Downloaded from http://ntr.oxfordjournals.org/ at University of North Dakota on July 8, 2015
and who is more likely to use it. Free and colleagues (2011), reporting on a trial of the txt2stop program in the United Kingdom, found that only a minority (38.5%) used the program’s “lapse” or “crave” functionality. Of those who reported at follow-up that they had lapsed, only 32% had reported their lapse when it occurred. Women and younger smokers were more likely to have done so (DeVries, Kenward, & Free, 2013). It is also important to understand user preferences for message frequency and duration. Free and colleagues (2009) found that some txt2stop users, who received 5 messages per day for 4 weeks after their quit day, found the program too “intense” or “intrusive,” yet others wanted the messaging to last longer. This suggests not only considerable acceptance of even quite high messaging schedules but also the desirability of being able to tailor message frequency. This paper reports on aspects of use of an interactive, automated text messaging advice program (QuitTxt, named Quit onQ at the time the data were collected), which sends tailored SMS messages several times daily until users either quit for 1 month, or have received messages for 20 consecutive days without having set a quit date. In the pragmatic trial from which these data are sourced, QuitTxt was found to be moderately effective, with efficacy related to overall usage (Borland et al., 2013). Imputed effect size was comparable to those of other programs (Whittaker et al., 2012). As trial participants offered QuitTxt were not obligated to use it, patterns of use are likely to be similar to those found under real-world conditions. QuitTxt encourages users to SMS changes in their status when they occur so that ongoing message content is appropriate, and “emergency help” is provided via code words corresponding with several types of tempting situation. In particular, we were interested whether patterns of use during the trial were consistent with the program’s design, and thus may have contributed to effectiveness. We examine use of the interactive capacity to report changes in quit status and to request emergency help, and use data from a follow-up assessment conducted 1 month after beginning use to estimate the extent of inadequate use, that is, failure to comply with reporting changes in quit status. We also examine preferences for messaging intensity (number of messages per day) and duration of use (including the extent to which the messages were manually stopped), and whether preferences can be predicted from user characteristics. Finally, we explore whether any preferences or patterns of use were associated with short-term cessation outcome or the perceived helpfulness of the program.
Smoking cessation text messaging intervention Table 1. Minimum Number of Status Change Messages for Use to Be Considered Adequate (Below This, Use Was Considered Inadequate) and Proportions Reporting Adequate Use, by Initial Progress to Quit Status at entry No quit attempts One quit attempt/relapsed One quit attempt/quit at follow-up 2+ quit attempts/relapsed 2+ quit attempts/quit at follow-up Reported all necessary status changes Missed reporting at least one change Did not make any status change Proportion needing to who reported all status changesa
No quit date
Quit date
Quit
0 2 1 4 3 30.5% 43.9% 25.6% 41.0%
1 2 1 4 3 38.0% 58.0% 3.9% 39.6%
NA 1 0 3 2 19.4% 42.9% 37.8% 31.0%
Note. NA = not applicable. aAmong those for whom there was evidence of a status change.
System-Derived Data Data on program interaction were derived from server log files, including duration of use (number of days received messages), number of messages received, how the program ended (i.e., whether the user sent stop or graduated), smoking status at graduation, number of status change and emergency help messages sent, and concurrent use of QuitCoach. Data Analysis Chi-square analysis was used to explore differences in user preferences and subsequent program interaction by baseline progress to quit, and their relationship to cessation outcomes and perceived helpfulness. Logistic regression was used to determine baseline predictors of program interaction; in addition to the baseline variables listed previously, the models included sample (help seeker or cold-contacted) and whether QuitTxt was used alone or integrated with QuitCoach.
Results Acceptance of Message Settings About a quarter (27.9%) chose to set different start or end times for weekends compared to weekdays. The majority (68.9%) chose to accept the default highest message intensity (3–8 messages per day), 26.2% chose the lowest message intensity (1–4 messages per day), while only 4.9% chose the medium message intensity (2–6 messages per day). Choice of medium or low message intensity was significantly more likely among those who had already stopped smoking at entry (Table 2). Nonetheless, even among this group, the majority (57.1%) opted for the highest intensity. Logistic regression (Table 3) confirmed this, also finding that lower nicotine dependence predicted choosing to reduce the message intensity.
1588
Program Use and Quitting Activity Almost all (93.3%) received messages for at least 4 days, which we have elsewhere defined as our criterion of use, based on an assessment of the minimum exposure needed to potentially derive a therapeutic benefit (Borland et al., 2013). From this point, we only consider the 862 participants who used QuitTxt for at least 4 days, most of whom (88.6%) were still smoking at entry. A minority (28.2%) also used QuitCoach in conjunction with QuitTxt, 42.9% of those for whom it was available. However, only 11.9% overall used QuitCoach during the period they were receiving messages, as most completed only one QuitCoach assessment, typically at the time of recruitment before or within minutes of the first QuitTxt message. One-month follow-up data were obtained from 93.5% of QuitTxt users. Of these, 23.2% did not make a quit attempt, 38.5% made a quit attempt but relapsed (includes those already quit at baseline), and 38.3% were abstinent at follow-up. Choosing a reduced message intensity was not associated with making a quit attempt by 1-month follow-up, but predicted greater abstinence among those who made a quit attempt or were already quit at baseline (Table 4). When stratifying by baseline smoking status, this remained significant in the same direction among baseline smokers (χ2(1) = 6.93, p = .008), but there was only a trend among recent quitters, possibly because of the much smaller size of this group (χ2(1) = 2.76, p = .097). No interaction was found (p = .60), suggesting the effect was similar in both groups. Just over half of users (52.2%) found QuitTxt to be very helpful, 34.9% somewhat helpful, and with only 12.9% not at all helpful. Baseline smokers who made a quit attempt were, as expected, more likely to have found the program helpful than those who did not make a quit attempt (χ2(1) = 7.48, p = .006), as were those who were abstinent at follow-up more likely to find it helpful than those who relapsed (χ2(1) = 11.94, p = .001). Logistic regression (Table 3) indicated that help seekers and those who also used QuitCoach were more likely, and women marginally more likely, to report QuitTxt to have been very helpful. Moreover, those who chose the highest intensity were more likely to rate QuitTxt “very helpful” (Table 4). Duration of Use Users received messages for a median of 27 days (interquartile range [IQR] 20–40), in that time receiving a median of
Downloaded from http://ntr.oxfordjournals.org/ at University of North Dakota on July 8, 2015
initial progress to quit (see Table 1). It should be noted that doing nothing was the appropriate action for those who began without a quit date and did not make an attempt, those who began with a quit date yet stopped the program before this quit date was reached, and those who began having already quit and did not relapse. Consequently, these groups were excluded from analyses of inadequate use.
Nicotine & Tobacco Research Table 2. QuitTxt Settings, Adherence, and Interaction, by Baseline Progress to Quit
Program activators Program settings Chose highest (default) message intensity Wanted different start or end time for weekends Program users Program adherence and interaction Reached termination (sent status change message) Reached termination (no status change message) Stopped prior to termination Sent a status change message Overall Made a quit attempta Did not make a quit attempta Sent an emergency help messageb
Smoker, no quit date
Smoker, set quit date
Already quit
Total
(n = 473)
(n = 346)
(n = 105)
(n = 924)
69.7% 28.4% (n = 441)
71.3% 25.6% (n = 321)
57.1% 33.7% (n = 100)
30.0% 44.6% 25.4%
35.8% 24.1% 40.2%
43.5% 58.0% 19.8% 28.6%
62.6% 70.9% 35.1% 24.9%
×2
p value
68.9% 27.9% (n = 862)
7.84 2.59
.02 .27
12.0% 55.0% 33.0%
30.0% 38.2% 31.8%
55.47
Original Investigation
How Do Smokers Use a Smoking Cessation Text Messaging Intervention? James Balmford PhD, Ron Borland PhD VicHealth Centre for Tobacco Control, Cancer Council Victoria, Melbourne, Australia Corresponding Author: James Balmford, PhD, VicHealth Centre for Tobacco Control, Cancer Council Victoria, 615 St. Kilda Road, Melbourne, VIC 3004, Australia. Telephone: 61-3-9514-6291; E-mail: [email protected] Received March 5, 2014; accepted May 31, 2014
Abstract
Methods: QuitTxt was offered during participation in a previously reported randomized controlled trial and was activated by 924 smokers or recent quitters, of whom 862 used it to a criterion level. Outcome data (quit attempts, 7-day point prevalence abstinence, and perceived helpfulness) were collected 1 month after first use. Results: Most (68.9%) accepted the default of 4–8 messages per day, and median use duration was 27 days. Half (49.1%) appeared to miss reporting at least 1 status change, with relapses less likely to be reported than quit progression. Emergency help was used by 27.0% of those eligible for it; emergency help was used more frequently among those with recent quit experience and lower nicotine dependence. Use of emergency help was unrelated to short-term cessation outcome. Conclusions: The most notable finding is the variability in use. Some users complied fully with the requirement to report status changes, while even among those who did not, many found QuitTxt to be very helpful, suggesting that perfect congruence between message content and quit status is not essential. The use of emergency help functionality was relatively rare but was appreciated.
Mobile-phone-based interventions of text (SMS) and/or video messaging are an effective method of delivering smoking cessation assistance, with a pooled risk ratio of 1.71 (95% CI = 1.47– 1.99) across five studies (Whittaker et al., 2012) including the text messaging study of ours that is elaborated on in this paper (Borland, Balmford, & Benda, 2013). These programs provide brief snippets of advice to help prompt action, strategies to resist temptations to smoke, timely reminders to prepare for upcoming challenges, and facilitate persistence in a quit attempt. A unique benefit is their ease of use anywhere at any time, while shared benefits with other automated programs include the capacity for tailoring, appeal to young people, and the potential for high reach at low cost-per-user. While the delivered content of these programs is specified, little is known about how these programs are actually used, including to what extent interactive components are utilized. To the extent that programs are not used as intended, their effectiveness may be reduced. Understanding how programs are used is necessary if we are to shape them to be both effective and maximally acceptable to users. In most published reports of texting programs (e.g., Free et al., 2011; Rodgers et al., 2005), the program requires a quit
date and provides messages assuming that the person quits on that date and stays stopped. Some recent programs (e.g., Haug, Meyer, Schorr, Bauer, & John, 2009; Snuggs et al., 2012) send periodic active requests for status information, thereby ensuring appropriate ongoing tailoring of content, while ours (Balmford, Borland, Benda, & Howard, 2013; Borland et al., 2013) requires sending specified text messages to notify of changes in situation (setting a date, initiating a quit attempt, and relapse). What little evidence there is suggests that smokers tend to comply with requests for status information, particularly if they have progressed in their quit attempt as opposed to relapsed (Snuggs et al., 2012). The timing of these requests is unlikely to correspond with the timing of actual changes, however, and it is not known to what extent users will report changes unprompted at the time they occur. Text messaging programs can also provide “emergency help” in response to user-initiated requests for support. This is a component of most mobile-phone-based interventions for smoking cessation, and potential users want and expect such functionality (Bock, Heron, Jennings, Magee, & Morrow, 2012). However, little is known about its perceived helpfulness
doi:10.1093/ntr/ntu111 Advance Access publication July 16, 2014 © The Author 2014. Published by Oxford University Press on behalf of the Society for Research on Nicotine and Tobacco. All rights reserved. For permissions, please e-mail: [email protected].
1586
Downloaded from http://ntr.oxfordjournals.org/ at University of North Dakota on July 8, 2015
Introduction: Mobile-phone-based text (SMS) messaging is an effective method for delivering smoking cessation assistance; however, little is known about optimal program use. This paper reports on the use of 2 forms of interaction (reporting changes in quit status and emergency help) among users of QuitTxt, an interactive, automated text messaging advice program. We examined preferences for messaging intensity, duration of use, and their associations with short-term cessation outcome or perceived helpfulness.
Nicotine & Tobacco Research
Methods Participants and Procedure The sample consisted of 924 smokers or recent quitters who registered with and subsequently activated the QuitTxt program to begin receiving messages during a study of “how effective Internet- and telephone-based resources are in helping smokers quit.” This represented 40.2% of the 2,299 participants from three arms of the study randomized to an offer of QuitTxt alone, or two ways of offering it in conjunction with the QuitCoach, an Internet-based expert system that provides tailored advice (Borland, Balmford, & Hunt, 2004). Study details, including outcomes, are provided in the study by Borland and colleagues (2013). Briefly, study participants were recruited from
help seekers (typically quitline callers, n = 869) or were coldcontacted from survey panels provided by a Melbourne-based market research company (n = 1,430). Over half (63.3%) of the 924 QuitTxt registrants came from the help-seeker group. Most of the help-seeker sample completed the recruitment assessment over the phone, whereas the entire cold-contacted sample responded to an E-mail invitation from the survey company and completed it online. A number of demographic- (age and gender) and smoking-related variables (nicotine dependence as measured by number of cigarettes per day and quit attempts in the past year) were measured at baseline, as was perceived comfort using SMS messaging. Progress to quit was categorized from standard questions on smoking status as no quit date set, quit date set, and already quit. QuitTxt Description and Signup QuitTxt is designed to be used by all smokers regardless of their current progress to quit and by recent quitters (up to 1-month quit). For those without a set quit date, the focus of messages is on motivating an attempt and gaining a commitment. For those with a quit date, QuitTxt is structured to lead the person to successful implementation, with a focus of activity in the day before the attempt. Once quit the focus moves to preventing relapse, initially by presenting strategies and motivational messages, but with more of an emphasis on achieving milestones over time. Three message intensities are available: default (3–8 messages per day), reduced (2–6 messages per day), and light (1–4 messages per day). Message intensity peaks around times of changes in status, for example, when one sets a quit date, around the quit date, and following relapse (if that happens), and then is gradually reduced. Tailoring of messages is primarily to progress to quit, with some other variables (e.g., age, gender, nicotine dependence) providing additional tailoring. The program is stopped at 1-month quit, if it is apparent that the user is not progressing (20 consecutive days without having set a quit date), or by request (by texting “STOP”). In the trial, all users were mailed a wallet-sized card with the commands recognized by QuitTxt to report changes in status, and mailed and/or E-mailed an A4-sized annotated version of the card. “Emergency help” is also offered to those who are quit, via several commands (e.g., STRESS, SOCIAL, BORED), with a coping response sent immediately, typically followed half an hour later by a message designed to encourage reflection on how well the strategy worked. Participants can also customize timing of the first and last messages of the day (messages are evenly spaced between these times), and whether they want different start and end times for weekdays and weekends. Follow-Up Data The 1-month follow-up, conducted via an E-mailed web link with phone follow-up for nonresponders, established whether participants had made a quit attempt since baseline, 7-day point prevalence abstinence, and other indicators of quitting activity. The 96.6% who reported evidence of quitting activity since baseline (a quit attempt, reduced consumption, or thoughts about quitting) were asked “How helpful was the program of text messages that you used?” and answered very helpful, somewhat helpful, and not at all helpful. Reported usage and status were combined with electronic records to estimate the extent of inadequate use; that is, failure to notify the program of status changes. The definition of inadequate use differed by
1587
Downloaded from http://ntr.oxfordjournals.org/ at University of North Dakota on July 8, 2015
and who is more likely to use it. Free and colleagues (2011), reporting on a trial of the txt2stop program in the United Kingdom, found that only a minority (38.5%) used the program’s “lapse” or “crave” functionality. Of those who reported at follow-up that they had lapsed, only 32% had reported their lapse when it occurred. Women and younger smokers were more likely to have done so (DeVries, Kenward, & Free, 2013). It is also important to understand user preferences for message frequency and duration. Free and colleagues (2009) found that some txt2stop users, who received 5 messages per day for 4 weeks after their quit day, found the program too “intense” or “intrusive,” yet others wanted the messaging to last longer. This suggests not only considerable acceptance of even quite high messaging schedules but also the desirability of being able to tailor message frequency. This paper reports on aspects of use of an interactive, automated text messaging advice program (QuitTxt, named Quit onQ at the time the data were collected), which sends tailored SMS messages several times daily until users either quit for 1 month, or have received messages for 20 consecutive days without having set a quit date. In the pragmatic trial from which these data are sourced, QuitTxt was found to be moderately effective, with efficacy related to overall usage (Borland et al., 2013). Imputed effect size was comparable to those of other programs (Whittaker et al., 2012). As trial participants offered QuitTxt were not obligated to use it, patterns of use are likely to be similar to those found under real-world conditions. QuitTxt encourages users to SMS changes in their status when they occur so that ongoing message content is appropriate, and “emergency help” is provided via code words corresponding with several types of tempting situation. In particular, we were interested whether patterns of use during the trial were consistent with the program’s design, and thus may have contributed to effectiveness. We examine use of the interactive capacity to report changes in quit status and to request emergency help, and use data from a follow-up assessment conducted 1 month after beginning use to estimate the extent of inadequate use, that is, failure to comply with reporting changes in quit status. We also examine preferences for messaging intensity (number of messages per day) and duration of use (including the extent to which the messages were manually stopped), and whether preferences can be predicted from user characteristics. Finally, we explore whether any preferences or patterns of use were associated with short-term cessation outcome or the perceived helpfulness of the program.
Smoking cessation text messaging intervention Table 1. Minimum Number of Status Change Messages for Use to Be Considered Adequate (Below This, Use Was Considered Inadequate) and Proportions Reporting Adequate Use, by Initial Progress to Quit Status at entry No quit attempts One quit attempt/relapsed One quit attempt/quit at follow-up 2+ quit attempts/relapsed 2+ quit attempts/quit at follow-up Reported all necessary status changes Missed reporting at least one change Did not make any status change Proportion needing to who reported all status changesa
No quit date
Quit date
Quit
0 2 1 4 3 30.5% 43.9% 25.6% 41.0%
1 2 1 4 3 38.0% 58.0% 3.9% 39.6%
NA 1 0 3 2 19.4% 42.9% 37.8% 31.0%
Note. NA = not applicable. aAmong those for whom there was evidence of a status change.
System-Derived Data Data on program interaction were derived from server log files, including duration of use (number of days received messages), number of messages received, how the program ended (i.e., whether the user sent stop or graduated), smoking status at graduation, number of status change and emergency help messages sent, and concurrent use of QuitCoach. Data Analysis Chi-square analysis was used to explore differences in user preferences and subsequent program interaction by baseline progress to quit, and their relationship to cessation outcomes and perceived helpfulness. Logistic regression was used to determine baseline predictors of program interaction; in addition to the baseline variables listed previously, the models included sample (help seeker or cold-contacted) and whether QuitTxt was used alone or integrated with QuitCoach.
Results Acceptance of Message Settings About a quarter (27.9%) chose to set different start or end times for weekends compared to weekdays. The majority (68.9%) chose to accept the default highest message intensity (3–8 messages per day), 26.2% chose the lowest message intensity (1–4 messages per day), while only 4.9% chose the medium message intensity (2–6 messages per day). Choice of medium or low message intensity was significantly more likely among those who had already stopped smoking at entry (Table 2). Nonetheless, even among this group, the majority (57.1%) opted for the highest intensity. Logistic regression (Table 3) confirmed this, also finding that lower nicotine dependence predicted choosing to reduce the message intensity.
1588
Program Use and Quitting Activity Almost all (93.3%) received messages for at least 4 days, which we have elsewhere defined as our criterion of use, based on an assessment of the minimum exposure needed to potentially derive a therapeutic benefit (Borland et al., 2013). From this point, we only consider the 862 participants who used QuitTxt for at least 4 days, most of whom (88.6%) were still smoking at entry. A minority (28.2%) also used QuitCoach in conjunction with QuitTxt, 42.9% of those for whom it was available. However, only 11.9% overall used QuitCoach during the period they were receiving messages, as most completed only one QuitCoach assessment, typically at the time of recruitment before or within minutes of the first QuitTxt message. One-month follow-up data were obtained from 93.5% of QuitTxt users. Of these, 23.2% did not make a quit attempt, 38.5% made a quit attempt but relapsed (includes those already quit at baseline), and 38.3% were abstinent at follow-up. Choosing a reduced message intensity was not associated with making a quit attempt by 1-month follow-up, but predicted greater abstinence among those who made a quit attempt or were already quit at baseline (Table 4). When stratifying by baseline smoking status, this remained significant in the same direction among baseline smokers (χ2(1) = 6.93, p = .008), but there was only a trend among recent quitters, possibly because of the much smaller size of this group (χ2(1) = 2.76, p = .097). No interaction was found (p = .60), suggesting the effect was similar in both groups. Just over half of users (52.2%) found QuitTxt to be very helpful, 34.9% somewhat helpful, and with only 12.9% not at all helpful. Baseline smokers who made a quit attempt were, as expected, more likely to have found the program helpful than those who did not make a quit attempt (χ2(1) = 7.48, p = .006), as were those who were abstinent at follow-up more likely to find it helpful than those who relapsed (χ2(1) = 11.94, p = .001). Logistic regression (Table 3) indicated that help seekers and those who also used QuitCoach were more likely, and women marginally more likely, to report QuitTxt to have been very helpful. Moreover, those who chose the highest intensity were more likely to rate QuitTxt “very helpful” (Table 4). Duration of Use Users received messages for a median of 27 days (interquartile range [IQR] 20–40), in that time receiving a median of
Downloaded from http://ntr.oxfordjournals.org/ at University of North Dakota on July 8, 2015
initial progress to quit (see Table 1). It should be noted that doing nothing was the appropriate action for those who began without a quit date and did not make an attempt, those who began with a quit date yet stopped the program before this quit date was reached, and those who began having already quit and did not relapse. Consequently, these groups were excluded from analyses of inadequate use.
Nicotine & Tobacco Research Table 2. QuitTxt Settings, Adherence, and Interaction, by Baseline Progress to Quit
Program activators Program settings Chose highest (default) message intensity Wanted different start or end time for weekends Program users Program adherence and interaction Reached termination (sent status change message) Reached termination (no status change message) Stopped prior to termination Sent a status change message Overall Made a quit attempta Did not make a quit attempta Sent an emergency help messageb
Smoker, no quit date
Smoker, set quit date
Already quit
Total
(n = 473)
(n = 346)
(n = 105)
(n = 924)
69.7% 28.4% (n = 441)
71.3% 25.6% (n = 321)
57.1% 33.7% (n = 100)
30.0% 44.6% 25.4%
35.8% 24.1% 40.2%
43.5% 58.0% 19.8% 28.6%
62.6% 70.9% 35.1% 24.9%
×2
p value
68.9% 27.9% (n = 862)
7.84 2.59
.02 .27
12.0% 55.0% 33.0%
30.0% 38.2% 31.8%
55.47
