Participants were recruited from the Houston metropolitan area using newspaper and billboard advertisements. A total of 1773 smokers were screened via telephone for study eligibility. Only healthy smokers who were interested in quitting smoking, had no uncontrolled medical illness, were fluent in English, and smoked at least 10 cigarettes per day were included in the study. Individuals were excluded if they were taking psychotropic medications, met the criteria for a current psychiatric disorder, or were involved in current smoking cessation activities. Of the 1773 smokers, 1035 (58.38%) met all the criteria and were invited to enroll in the study. However, of the 1035 smokers, 119 (11.5%) did not attend the baseline visit, resulting in 916 (88.5%) participants being randomized and issued a study device. Of the 916 randomized participants, 96 (10.5%) returned the device unused and thus were not exposed to treatment and were therefore excluded from the outcome analyses.

Participants could earn up to US $115 in gift certificates by completing study procedures. All participants received a US $50 gift certificate for the completion of visits through the end-of-treatment visit and a US $15 gift certificate for completion of the 2 follow-up visits. Participants also received a total of US $5 to US $50 in gift certificates based on completion of random mood and smoking assessments generated by their computer while in the study.

The protocol was approved by the institutional review board of The University of Texas MD Anderson Cancer Center. This trial was not registered because the data collection began before July 1, 2005. Data collection for this study was conducted from 2000 to 2006, before the advent of smartphones. The data analysis for this study began in 2018. The data from this study and the customized program provided the basis for a new scheduled smoking smartphone app designed to run on the current Android platform. This program is currently being tested and will be the subject of a future study. A clinical version (nonresearch) will be made available for download following testing.

Participants were randomly assigned to 1 of the following 3 groups: scheduled smoking plus a precessation nicotine patch (SSNP), scheduled smoking only with no precessation patch (SS), and enhanced usual care (EUC) control. The groups were stratified by sex, race, cigarettes smoked per day, and depression history. The study was divided into 4 phases: baseline (2 days), precessation date intervention (21 days; SSNP and SS groups only), postquit date intervention phase 1 (14 days; including the quit date), and postquit date intervention phase 2 (8 weeks). During the baseline period, all groups were instructed to smoke ad libitum and use a handheld device (HD) to record their smoking, and to receive other programming, as described in the section below. During the precessation period, both scheduled smoking groups (SSNP or SS) were exposed to a 21-day computerized intervention delivered on the HD, designed to progressively reduce daily cigarette consumption before their quit date. The SSNP group also began using a nicotine patch simultaneously with the reduction schedule. The EUC group did not receive a precessation patch or cigarette reduction intervention. During the postquit date intervention (phase 1), which included the quit date, all groups used the nicotine patch and HD to continue recording smoking behavior and receive other programming, as noted in the Scheduled Smoking section below. During the postquit date intervention (phase 2), all groups stopped using the HD but continued using the nicotine patch for an additional 8 weeks.

During the 3-week precessation period, smokers in both the SSNP and SS groups were instructed to smoke at scheduled times only. The computer-assisted scheduled smoking intervention application, active on the HD for the SSNP and SS groups only, prompted the smoker to smoke only at specific times during the day. Over a period of 3 weeks, the program gradually increased the interval between prompted cigarettes (ie, the intercigarette interval) such that only 3 to 4 cigarettes were scheduled to be consumed the day before the quit date. The reduction schedule was implemented using the following algorithm: (1) using the daily smoking mean and wake time derived from the participants’ self-report at baseline, the program calculated an intercigarette interval by dividing the average number of minutes awake per day by the reduced percentage of the smoker’s average daily smoking frequency. (2) For days 1 to 12, the baseline smoking rate was reduced by 15% every 3 days, and for days 13 to 21, a 10% reduction every 3 days was used (this process typically resulted in a 75% reduction in the baseline smoking rate by the beginning of the third week of the schedule and corresponding progressively longer intercigarette intervals). (3) The process outlined in step 2 was used to determine the maximum allowable cigarettes for that period; however, a 15-minute delay to the first cigarette of the day was added to the schedule each week, and because of this delay, any cigarette consumption that would have occurred past the participants’ bedtime were not added to the schedule. Moreover, once the smoking frequency was reduced to 3 to 4 cigarettes per day, no further reductions took place until the quit date, when zero cigarettes were expected. (4) All smoking times were adjusted to fall on the nearest quarter within the hour. A sample schedule over the 3-week scheduled smoking period is provided in Figure S1 in Multimedia Appendix 1 [1-22,29-58].

Participants in the SSNP and SS groups attended 9 in-person visits as follows: baseline, 3 weeks precessation, 2 weeks precessation, 1 week precessation, 2 days after quitting, 2 weeks after quitting, 4 weeks after quitting, 6 months after quitting, and 12 months after quitting.

Participants in the EUC group had no precessation scheduled smoking or reduction intervention and hence attended only 6 in-person visits (baseline, 2 days after quitting, 2 weeks after quitting, 4 weeks after quitting, 6 months after quitting, and 12 months after quitting).

At the baseline visit (approximately 45 minutes), participants were instructed on the use of the HD, which included watching a short video explaining the group-specific and general features of the program, and were assessed on their understanding of device use. All other visits (approximately 20 minutes) were used to obtain questionnaire information, assess abstinence, and download data from the HD. No smoking cessation counseling was provided during these visits.

During the 21-day precessation date intervention period (Figure S2 in Multimedia Appendix 1), smokers in the SSNP group began wearing a 14-mg nicotine patch on the first day of scheduled smoking, which was continued during the 3-week precessation period. Smokers in the SS group did not receive nicotine patches during the precessation period. During the postquit date period (phase 1), smokers in both the SSNP and SS groups began using the 21-mg patch on their quit date regardless of whether they quit early or not. Smokers in the EUC group began using the active 21-mg patch on the quit day, which occurred 2 days after their baseline visit.

During the 10 weeks following the target quit date, all the participants continued to use the nicotine patch. The participants used the 21-mg patch for 6 weeks, followed by the 14-mg patch for 2 weeks, and finally a 7-mg patch for 2 weeks.

The study device was the Cassiopeia Pocket PC E-125 (Casio Computer Co Ltd), an HD that was a precursor to today’s smartphones (Figure S3 in Multimedia Appendix 1). The device was used to deliver the custom programmed scheduled smoking intervention in the 2 scheduled smoking groups, along with assessments and routines common to all groups. To provide a control for HD use and support internal validity, those in the EUC group were issued the same HD with all the functionality provided to the scheduled smoking groups but without the scheduled smoking routine. The provision of the HD to the control group distinguished this group from a “usual care control” because the programming provided some “enhanced” features unrelated to scheduled smoking, which were designed to provide cessation tips, conduct smoking assessments, foster relapse prevention, and renew quit attempts following a relapse (refer to HD functionality description and Figure S3 both in Multimedia Appendix 1).

Other applications on the device were disabled. All groups received the study device on the day of randomization to treatment. The 2 scheduled groups used the device for a total of 37 days, including 2 days before the quit date (precessation), 3 weeks precessation, and 2 weeks after cessation (from the initial quit date). The EUC group used the device for 2 days precessation and 2 weeks after cessation.

The HD shared common functionality across groups, including an embedded electronic cessation tip guide; programmed assessments of smoking behavior, mood, and other conditions delivered at both fixed (daily diary) and random times (ecological momentary assessments); and user-initiated recording of actual smoking, temptations (urges) to smoke, and quit attempts. The wake-up time and bedtime were recorded/confirmed daily (Figure S2 in Multimedia Appendix 1). The tip guide consisted of a series of hyperlinks embedded in the program. Each smoker also received a printed version of the tip guide at randomization (Figure S4 in Multimedia Appendix 1).

The schedule smoking was initiated when the HD signaled a smoking interval by flashing a display and sounding an alarm. A sample screenshot for the first day of scheduled smoking is shown in Figure S5 in Multimedia Appendix 1. The program included routines (Figure S6 in Multimedia Appendix 1) to allow for early quitting in the scheduled groups (ie, making a quit attempt before the quit date). During the 2 weeks immediately following the quit date, participants in all 3 groups were provided with a relapse management routine (Figure S7 in Multimedia Appendix 1) if they indicated that they had smoked (even a puff) on the previous day during the wake-up routine or entered that they smoked a cigarette at any time (Multimedia Appendix 1 provides additional details on methods and device).

Written informed consent was obtained from each participant at screening. All participants received a US $50 gift certificate for completion of visits through the end-of-treatment visit and a US $15 gift certificate for the completion of the follow-up visits. Participants also received a total of US $5 to US $50 gift certificates based on completion of ecological momentary assessments. This research was approved by MD Anderson’s institutional review board (PA15-0675).

As shown in Figure 1, a total of 1773 smokers were screened for the study; 738 were excluded based on study exclusion criteria; 1035 were invited to enroll; and 916 presented at the baseline for randomization, with 306, 309, and 301 participants allocated to the SSNP, SS, and EUC groups, respectively. However, 33, 41, and 22 participants in each of the respective groups returned their HD unused and were excluded from the analysis for not having been exposed to the intervention.

The baseline demographic characteristics for the 3 main treatment groups (SSNP, SS, and EUC) and the 4 compliance groups (SSNP compliant, SSNP noncompliant, SS compliant, and SS noncompliant) are presented in Table 1. Typical for smoking cessation trials, participants had an average age of approximately 43 (SD 11.02) years, were mostly White, were equally distributed between male and female, smoked slightly more than a pack of cigarette per day (mean 23, SD 10.33), had been smoking for an average of approximately 24 years, and were moderately dependent on nicotine (mean Fagerström Test for Cigarette Dependence score 5, SD 1.98). The Center for Epidemiologic Studies Depression Scale depression total score and the Positive and Negative Affect Scale scores were consistent across the 3 treatment groups. No significant differences were found between the treatment and EUC groups on any of these baseline characteristics.

Collectively, participants in the SSNP and SS groups did not differ from each other in the percentage of scheduled cigarettes smoked (SSNP: 76% and SS: 74%; F_1,487=0.61; P=.43) or the number of off-scheduled cigarettes smoked (SSNP: 24 and SS: 25; F_1,487=0.31; P=.58) during the 3 precessation week (reduction) period.

Significantly more participants from the SSNP group than those in the SS group achieved ≥50% reduction in expired CO (Wald X²₅₂₁=5.5; P=.02) during the precessation phase. Specifically, 50.9% (139/273) of all participants assigned to the SSNP group achieved ≥50% reduction in expired CO level between their baseline and last precessation visit, whereas only 26.5% (71/268) of those assigned to the SS group reduced their expired CO level by ≥50%. Smokers in the SSNP group who achieved ≥50% CO reduction were more likely than those who failed to reduce their CO by 50% to have smoked their scheduled cigarettes on time (SSNP compliant: 80% and SSNP noncompliant: 71%; t_1,521=5.34; P<.001). However, the 2 groups did not differ in the number of cigarettes smoked off schedule (unscheduled cigarettes) during the 3-week precessation period (SSNP compliant: 26 and SSNP noncompliant: 22; t_1,521=0.978; P=.33). Similarly, smokers in the SS group who reduced their CO by ≥50% were more likely than those in the SS group who failed to reduce their CO by 50% to smoke on schedule (SS compliant: 78% and SS noncompliant: 72%; t₉₀₃=2.56; P=.01). The 2 groups did not differ in the number of unscheduled cigarettes smoked during the 3-week precessation period (SS compliant: 25 and SS noncompliant: 25; t₉₀₃=0.09; P=.92).

Participants assigned to the SSNP group reported a high rate of compliance to precessation use of the nicotine patch, applying an average of 97% (SD 10.7%) of all patches as instructed before their target quit date. Prequit nicotine patch compliance for SSNP compliant was 97% and for SSNP noncompliant was 95%. Postquit nicotine patch use was also assessed, and we found no significant difference in compliance across all groups at 2 weeks (SSNP: 90.96%, SS: 89.25%, and EUC: 90.72%; F_2,626=0.37; P=.69) and 4 weeks (SSNP: 81.88%, SS: 82.22%, and EUC: 80.20%; F_2,566=0.22; P=.80) after the target quit date. Across the 4 compliance groups, use of nicotine patch was also similar with no significant differences for the 2 weeks postquit (SSNP compliant: 91.43%, SS compliant: 92.62%, SSNP noncompliant: 89.56%, and SS noncompliant: 87.4%; F_3,504=0.4; P=.75) as well as for the 4 weeks postquit time points (SSNP compliant: 86.10%, SS compliant: 83.01%, SSNP noncompliant: 73.81%, and SS noncompliant: 81.57%; F_3,367=1.93; P=.12).

From a design perspective, this high level of patch compliance means that after the quit date, the 3 treatment groups (SSNP, SS, and EUC) were essentially the same, that is, scheduled smoking was completed, and all were now equally exposed to a standard regimen of NRT.

This innovative clinical trial evaluated the effects of a mobile smoking cessation intervention designed to progressively reduce smoking before quitting using SSNP, in comparison with SS and an EUC control group. EUC smokers used the nicotine patch on the quit date and carried the same mobile device to record cigarettes and manage their quit attempt, with no scheduled reduction. No counseling was provided in this study beyond the explanation of the use and rationale of the procedures and an electronic tip guide available on the mobile device. The results showed no overall differences in abstinence in the 3 groups over multiple time points (2 weeks, 4 weeks, 6 months, and 12 months posttarget quit date) in both unadjusted and adjusted (for covariates) analyses using both ITT (missing smoking status imputed as smoking) and RO (no imputation) approaches to the treatment of missing data. However, when taking schedule compliance into account, a different pattern of results emerged.

We evaluated schedule compliance in 2 types of analyses, and the results in both were consistent, showing that smokers who were compliant with the scheduled smoking procedure and used the nicotine patch before the quit date were more likely to be abstinent at several time points, with the differences at 2 weeks and 4 week after quitting (primary outcome of the study) being the most robust.

Specifically, when we subdivided the 2 scheduled smoking groups into those who complied with the scheduled reduction, as defined by a 50% drop in expired CO between baseline and final week of scheduled smoking, and those who did not, we found that the SSNP compliers abstained significantly more often at the 2-week, 4-week, and 6-month postquit dates when compared with the EUC group. Although the differences at 4 weeks and 6 months were not significant following the correction for multiple comparisons, they continued to trend in the correct direction, with ORs >1.5, which have generally been considered as a clinically meaningful difference between treatments.

We also took an IV approach to evaluate compliance, using LATE estimates of the direct causal effect of the smoking cessation intervention among compliers. In comparison with the EUC group, those who complied with the intervention in the SSNP group averaged a 26%, 29%, and 22% increase in abstinence at the 2-week, 4-week, and 6-month postquit time points, respectively, although the 6-month outcome was not statistically different. Thus, this LATE analysis offers support for the value of the SSNP intervention when compliance is considered. The LATE analysis complements the standard effectiveness analyses because the LATE analysis evaluates the effect of actually receiving the intervention. We found no differences across the 3 groups in postquit compliance with patch use. However, precessation patch use was related to schedule compliance, as noted by the fact that >50% of the smokers in the SSNP group achieved a >50% reduction in CO (a criterion of compliance) compared with 26% in the SS group. However, both the direction and the magnitude of the effects of SSNP schedule compliance in both compliance analyses gives us confidence that compliance with the scheduled smoking procedure is a key factor for the intervention to work. This conclusion is further supported by the absence of precessation differences in compliance with the nicotine patch between the 2 SSNP compliance groups (SSNP compliant: 97% and SSNP noncompliant: 95%). Thus, although use of the nicotine patch before cessation may make it easier for smokers to comply with the scheduled reduction procedure, compliance with precessation patch use alone is insufficient to improve smoking cessation outcomes over usual care. Moreover, we noted that the SSNP intervention had other beneficial effects on the quitting experience, such as reducing withdrawal, negative affect, and craving, which could reduce the probability of relapse, and may encourage future quit attempts should that occur, particularly because we know that it may take multiple quit attempts before a smoker eventually succeeds [8]. Subsequent studies should evaluate ways to promote compliance, notably through smoking cessation counseling, which was not conducted in this study.

As noted earlier, previous studies on gradual reduction methods have been largely inconclusive when compared with abrupt quitting (usual care); however, the efficacy of precessation nicotine patch use has received modest support. Our results suggest that the combined use of nicotine patch and scheduled smoking (a method of gradual reduction) can be effective when compliance with the scheduled smoking procedure is observed. This is consistent with a study by Rose et al [71] that showed that the beneficial effects of precessation patch use on smoking cessation are strongly related to the extent of precessation smoking reduction as measured by changes in expired CO from baseline to the quit date, which we used as a criterion for assessing schedule compliance. Our results are also consistent with previous studies suggesting that scheduled smoking has positive effects on the quitting experience, reducing withdrawal, negative affect, and craving.

Although the overall comparisons showed no differences across the groups, there are limitations in the study design that should be noted. First, the main component of this intervention was implemented before the quit date and involved simultaneous use of both the nicotine patch and the scheduled smoking procedure. However, the postquit intervention was essentially the same for all groups and consisted of continued use of the HD for an additional 2 weeks plus use of the nicotine patch. Thus, the expectation that a 3-week prequit intervention might have long-lasting effects on postquitting behavior would require a very large precessation treatment effect, when averaging across all participants within a group. Our compliance analyses suggest that the effects of the combined scheduled smoking and nicotine patch can persist into the postquitting environment, with clinically meaningful effects up to 6 months after quitting among smokers who adhere to the procedure. This is analogous to the notion that a drug cannot work if it is not taken, and a key factor is to determine why it is not used and to intervene if possible. This intervention by design did not involve smoking cessation counseling, which we know is essential to produce improved outcomes when combined with pharmacotherapy [35]. Although we envisioned an intervention that would function autonomously without counseling, the program could have been enhanced by incorporating modern text messaging, chatbots, or artificial intelligence components to monitor and enhance compliance as well as provide limited motivational counseling. These features are incorporated into our current scheduled smoking program. Moreover, providing the SSNP intervention in the context of traditional smoking cessation counseling may be an important implementation strategy. Such an approach could easily be incorporated into state quitline interventions that involve text message interventions. Quitline counseling could provide counseling and monitor adherence.

Another limitation was the provision of HD in the control group. This was done to achieve internal validity, that is, all smokers carried a device that delivered the scheduled smoking intervention to both scheduled groups in the precessation period and functioned essentially the same for all smokers in the postquit environment. Thus, the control group had access to the relapse management routines, tip guides, and self-assessments, all of which might have had an impact on postquit date smoking behavior and raising cessation rates beyond what might be expected through usual care (ie, quitting abruptly, with the nicotine patch). None of these features would normally be available under true usual care control groups; hence, this group was designated as “Enhanced Usual Care.”

This study evaluated the combination of a low-touch behavioral treatment (scheduled smoking) delivered on a HD with and without NRT against a usual care NRT control condition. It could be argued that because our findings and program used to produce the data were >20 years old, they may have limited applicability to the current treatment standards, technology, and the current population of smokers. However, we believe that our findings are highly relevant to the current smoking cessation treatment for several reasons. First, the basic standard of care for smoking cessation till date is NRT, typically using the nicotine patch [10]. It is by far the most used pharmacotherapy with quitlines [36] and low-touch interventions such this study. In addition, as described earlier, gradual reduction methods remain attractive to many smokers trying to quit; hence, both the behavioral (scheduled smoking) and pharmacological (NRT) treatments fit into today’s approaches to smoking cessation. Second, although smoking rates have declined nationally from approximately 23% in 2002 [37] to 12.5% in 2020 [38], our inclusion criteria for this study included smoking ≥10 cigarettes smoked per day at baseline, which captures most smokers (approximately 75% in 2015 [39]—latest available data). Third, with regard to the technology used to deliver the intervention, the hardware device is obsolete, but as indicated earlier, we have adapted (and enhanced) each of the elements of the program for use on currently available smartphones (Android) and have been testing the program in a clinical setting. We intend to make the program available to the research community and include an iOS version when complete.

Scheduled smoking, when combined with precessation use of NRT, can result in significantly higher abstinence rates than usual care (abrupt cessation with NRT), particularly in the early postquit phase (2 and 4 weeks after cessation) when smokers are compliant with the procedure. There was also evidence for a clinically meaningful effect (OR>1.5) at 6 months after quitting, although statistical differences were reduced with adjustments for multiple comparisons. Scheduled smoking also produces a better overall quitting experience by reducing symptoms of nicotine withdrawal and craving in comparison with usual care, which could encourage more future quit attempts. Future studies in this area should focus on the use of counseling or other methods to improve adherence.