A Secondary Analysis of a Preliminary Contingency Management Intervention for Presurgical Cancer Patients: Evaluating Individual Participant Data

Abstract

Introduction

Contingency management (CM) interventions deliver monetary reinforcers contingent upon biochemically verified abstinence from smoking. CM has been found to be effective, however, individual participant, analyses are warranted to further elucidate how individual-level behavior patterns vary during the intervention period, both within and across treatment groups.

Aims and Methods

This is a secondary analysis of a randomized controlled pilot trial of presurgical cancer patients who smoke (RCT N = 40). All participants were current everyday smokers and were enrolled in cessation counseling, offered nicotine replacement therapy, and submitted breath CO testing 3 times per week for 2–5 weeks. Participants randomized to CM received monetary reinforcers for breath CO ≤6 ppm on an escalating schedule of reinforcement with a reset for positive samples. Sufficient breath CO data exist for 28 participants (CM = 14; monitoring only [MO] = 14). Effect size was calculated for differences in negative CO tests. Time to first negative test was tested using survival analysis. Fisher’s exact test was used to assess relapse.

Results

The CM group achieved abstinence more quickly (p < .05), had a lower percentage of positive tests (h = 0.80), and experienced fewer lapses following abstinence (p = .00). While 11 of 14 participants in the CM group achieved and sustained abstinence by their third breath test, this was only true for 2 of the 14 MO participants.

Conclusions

Those in CM achieved abstinence quicker and with fewer lapses than those engaged in MO speaking to the efficacy of the schedule of financial reinforcement. This is particularly important within presurgical populations given the potential benefits to postoperative cardiovascular and wound infection risk.

Implications

While the efficacy of CM as an intervention is well established, this secondary analysis provides insight into the individual behavior patterns underlying successful abstinence. Those assigned to CM were not only more likely to achieve abstinence, but did so more quickly and with fewer instances of relapse. This is of particular importance to those scheduled for surgery where achieving abstinence as early as possible impacts on the risk of postoperative complications. CM interventions may be particularly well suited for critical windows in which timely and sustained abstinence is advantageous.

Introduction

In contingency management (CM) interventions for tobacco cessation, financial incentives are delivered contingent on biochemically verified abstinence (eg, breath CO at or below a specified target amount). CM interventions have been found to be highly effective in producing cessation of smoking and substance abuse^1,2 as well as improving chronic disease management behaviors.³ While CM efficacy has been demonstrated in randomized clinical trials, analyzing individual participant data to identify the varied behavior patterns underlying group-level findings are warranted to further elucidate how individual-level behavior patterns change during the intervention period within and across groups.

Such an idiographic approach may illuminate the presence of individual variation in intervention efficacy and functionally distinct subgroups that may be otherwise obscured in group-level analyses.⁴ That is, while individuals in both the treatment and control groups will demonstrate abstinence at treatment completion, the behavior patterns underlying their quit attempt may differ as a function of the intervention they receive. For example, visual inspection of data reported in previous research indicates a large proportion of individuals within CM interventions achieve abstinence soon after the initiation of the intervention (~50%–75%), a small proportion of individuals obtain abstinence after a delay (5%–7%), and the remaining participants make little or no contact with the reinforcement contingency (25%–35%).^5–7 Achieving abstinence in a short time frame is particularly important in presurgical cancer populations as the duration of abstinence before undergoing a surgical procedure impacts the risk of postoperative complications.⁸ Those who quit smoking at least 4 weeks before surgery have been shown to experience fewer respiratory complication⁹ while those who achieve abstinence at least 3 weeks before surgery experience fewer wound healing complications.^9,10 Thus, while an individual participant may have met abstinence criteria on the day of their surgery in the primary study (7-day point prevalence abstinence), longer durations of sustained abstinence from smoking are beneficial for this specific population.¹¹

By characterizing the behavior patterns of responders and non-responders to CM interventions, we may be able to develop structured approaches to better identify individuals who are less likely to achieve abstinence early in the course of treatment and tailor or intensify interventions as needed. The present secondary analysis of individual breath carbon monoxide data in the context of the first randomized CM trial of preoperative patients seeks to characterize such patterns.

Methods

This is a secondary analysis of a preliminary randomized controlled pilot trial of CM among presurgical cancer patients who smoke, with a complete description of the methodology presented in the primary publication.¹¹ Data for this trial were collected at Yale Cancer Center in New Haven, CT, and the MUSC Hollings Cancer Center in Charleston, SC. This project was approved by the Medical University of South Carolina Institutional Review Board (#00054733) and the Yale University Human Investigation Committee (#1407014258). Clinical trial registration number: NCT02402023.

Participants

Participants were current everyday adult (age ≥18 years old) smokers (≥ 1 cigarette per day at baseline) diagnosed or suspected to have cancer and scheduled for surgery between 10 days and 5 weeks from the time of enrollment. Exclusion criteria included: Unstable medical or psychiatric conditions such as suicidal ideation, acute psychosis, severe alcohol dependence, or dementia. A total of 40 participants were consented and randomized.

Nicotine Replacement Therapy and Cessation Counseling

All participants received treatment in the form of both nicotine replacement therapy and tobacco cessation counseling. nicotine replacement therapy was provided in the form of patches based on baseline tobacco intake (21 mg for >10 cigarettes per day and 14 mg for ≤10 cigarettes per day). Patches were provided regularly until the date of their surgery. Short-acting nicotine replacement therapy was not administered in this pilot trial in order to best assess the contribution of CM procedures on abstinence rates. All participants also engaged in regular tobacco cessation counseling with a tobacco treatment program provider (advanced practice registered nurse, clinical pharmacist, or psychologist). Participants were asked to set a quit date within one week of their initial visit and received 2–5 manualized cognitive behavioral counseling sessions.¹²

Breath CO Testing

All participants were required to submit breath CO tests 3 times per week for the duration of the trial, resulting in 6–16 total tests. All participants had the opportunity to view their breath CO reading at each timepoint. Treatment duration was variable across participants due to scheduling variability prior to surgery. CO levels ≤6 parts per million (ppm) were considered negative for smoking. Participants were also asked to self-report on abstinence at each testing period. The monitoring only (MO) condition received no financial incentives based on the result of breath tests.

In addition to the treatment outlined above, those in the CM group also received monetary reinforcers contingent upon biochemically verified abstinence. The schedule of reinforcement started at $15 for the first negative test and increased by $5 for each subsequent negative test, with a ceiling of $55 (ie, for participants who earned $55 per abstinent test, their payment remained at this level for the duration of their confirmed abstinence testing). In the case of either missing tests or tests resulting in CO >6 ppm, participants did not receive compensation and the incentive schedule was reset, resuming at $15 on the next negative test. CM delivery spanned the weeks between consent and surgery (10 days-5 weeks).

Data Analysis

In order to visually inspect individual pathways of cessation over the course of the treatment, each breath CO test was plotted and arranged by group. Effect sizes were calculated for: time to first negative CO test, and percentage of negative CO tests. A Fisher’s exact test was used in order to assess the relative risk for a positive CO breath test following a negative one. In order to examine between group differences in time to first abstinence a Kaplan–Meier survival analysis was completed utilizing a log-rank test of significance. A series of independent T-tests and chi-squared analyses were utilized to determine if any demographic variables, nicotine dependence as measured by the Fagerstrom Test for Nicotine Dependence, and readiness to quit as measured by a 0–100 rating, were associated with response to treatment.

Results

A total of 14 participants in the CM condition and 14 participants in the MO provided sufficient data to be graphed (n ≥3 CO tests). The remaining 12 participants (MO = 5; CM = 7) submitted ≤2 CO tests, because of withdrawal from the study or loss to follow-up and are not included in the individual analysis. Demographic (female = 53.6%; white = 64.3%; mean age = 37 SD = 9.58) nicotine dependence (m = 4.36, SD = 2.34), and readiness to quit (m = 73.10 SD = 28.00), data did not vary between groups. Individual CO graphs are presented in Figure 1.

Figure 1.

Breath carbon monoxide by testing period. Individual plots are breath carbon monoxide (CO) in parts per million (ppm) at each testing period. Variation in the length of the X axis reflects variation in time-to-surgery across participants. The first vertical dashed line indicates the onset of intervention, and the second indicates the break between surgery and the 3-month follow-up. The horizontal dotted line indicates the 6ppm cutoff for abstinence.

Survival analysis indicates that those in the CM condition achieved their first negative breath test more quickly than those in the MO condition with 77% of those in the CM condition achieving abstinence by time 2 compared to 20% in the MO group (p < .05). The survival plot is presented in Figure 2.

Figure 2.

Kaplan–Meier survival plot for time to first abstinence.

Additionally, throughout the course of testing, those in the CM condition provided considerably higher percentages of negative tests (85.7% vs. 49.6%; h = 0.80). CM participants were also considerably less likely to submit a positive test following a negative test (CM 0.9% (1/108) vs. Monitoring 20.4% (10/49); p < .001). While 11 of the 14 participants in the CM group achieved abstinence without subsequent lapse by their third breath test, this was only true for 2 of the 14 MO participants. Additionally, 9 of the 10 participants in the CM group who achieved abstinence on a single test maintained that abstinence for the remainder of the trial, but this was true for only 4 of the 10 in the MO condition. All participants who successfully achieved abstinence in the CM intervention produced a negative breath test by the fifth testing period. For those who achieved abstinence at the time of surgery across both groups, 90.5% produced their first negative test by the third breath test; (ie, within 6–9 days). In addition, among those in the CM condition, approximately 70% of participants who were abstinent at the time of surgery maintained their abstinence at the 3-month follow-up. In the MO condition, only 25% of those abstinent at surgery maintained their abstinence at follow-up. Among those randomized to CM, and using an intent-to-treat approach with all 21 participants, 57.1% achieved abstinence by the fifth testing period, and 42.9% could be characterized as non-responders (ie, they never contacted the abstinence contingency; seven of whom were due to dropout). Within the MO condition, 36.8% of individuals were early responders, achieving abstinence by the fifth testing period (ie, within 9–11 days), 15.8% were delayed responders, and 47.4% as non-responders (five of whom were due to dropout). Table 1 presents the response to treatment by group. Demographics including age (t = 1.95; p = .65), gender (X² = .02; p = .88), and race (X² = 1.05; p = .31), as well as nicotine dependence (t = −1.18; p = .21), and readiness to quit (t = 1.27; p = .45) were not found to be significantly different between early and delayed or non-responders.

Table 1.

Response to Treatment by Group

Group assignment	N	Early responder (≤5 tests)	Delayed responder (>5 tests)	Non-responder (no negative tests)
Contingency management	21	12(57.1%)	0(0%)	9(42.9%)
Monitoring only	19	7(36.8%)	3(15.8%)	9(47.4%)

Discussion

Presurgical, smoking cancer patients assigned to CM achieved abstinence more quickly and with fewer lapses than those engaged in MO. This speaks to the efficacy of financial reinforcement within the CM intervention to not only facilitate early abstinence but to also maintain it. This is especially important as presurgical smoking cessation significantly reduces the risk of postsurgical complications in cancer patients.^13,14 The intensity and duration of presurgical smoking cessation interventions have shown to be critical in reducing postsurgical complications¹⁵ which is especially notable given the brief period between cancer diagnosis and surgery. Half of the CM participants achieved abstinence at least 3 weeks before their scheduled surgery, and three achieved abstinence at least 4 weeks before surgery. These presurgical quitting milestones are clinically important, as they are associated with lower risk for wound healing complications^9,10 and fewer postoperative respiratory complications.⁹ In contrast, early and sustained abstinence was evident in only one MO participant.

Previous research has found that the duration of smoking abstinence is associated with diminished reinforcing effects of cigarettes,^16,17 reducing the likelihood of subsequent relapse. The present data are also consistent with findings that CM interventions can effectively induce abstinence even after an initial lapse.¹⁸ Whether the CM patterns found herein are specific to this study and its schedule and magnitude of contingencies, or whether these patterns are more common across a range of CM interventions and populations is unclear. Future studies on CM interventions should consider conducting similar individual-level analyses to characterize behavior patterns and potentially replicate these findings.

These findings also provide insight into how CM interventions can be adapted depending on treatment response. Across conditions, an individual’s treatment response can be determined relatively early into treatment, presenting an opportunity to redirect treatment efforts for individuals who have not achieved abstinence. Those who did not respond to the intervention may have required a greater magnitude of financial incentives for negative tests,¹⁹ had a higher valuation of tobacco, or more steeply discounted greater distal (vs. smaller immediate) rewards.²⁰ Further research should test the need for more intensive interventions for CM non-responders within the context of adaptive designs which might reallocate patients based on their response to initial treatment, allowing researchers to test more intensive or tailored interventions specifically for non-responders.²¹

There are several limitations to this analyses that should be considered. The intervention period of this study was determined by an individual’s scheduled cancer surgery which necessarily resulted in brief, uneven intervention durations. Given a longer observation period, it is possible that some additional lapses may have been observed. Secondly, this trial utilized a relatively thin schedule of CO testing which may have made brief lapses harder to detect. The sample size analyzed in this preliminary analysis is also small, has a considerable attrition rate, and contains limited available follow-up data. These findings should be replicated in a larger trial. Another potential concern is in regards to the feasibility of implementing CM interventions given the cost.²² Possible suggested methods for paying for CM interventions include “deposit contracts” (patients depositing monetary incentives at the start of treatment to earn back), coverage by an insurance company for whom the benefits of CM may be cost-effective, healthcare systems with established populations such as the United States Department of Veteran Affairs,²³ or solicitation of community donations. For this population specifically, the reduced risk of postsurgical complications and readmission may render CM interventions cost-effective for hospital systems, especially when treating uninsured patients.

In sum, while the efficacy of CM as an intervention is well established, this secondary analysis provides insight into the individual behavior patterns underlying successful abstinence. Those assigned to CM were not only more likely to achieve abstinence, but did so more quickly and with fewer instances of relapse. This is of particular importance to those scheduled for surgery where achieving abstinence as early as possible impacts on the risk of postoperative complications. CM interventions may be particularly well suited for critical windows in which timely and sustained abstinence is advantageous.

Funding

This work was supported by the National Cancer Institute at the National Institutes of Health grants (R21CA181569, R01CA235697 to BAT; K07CA214839 to AMR; NIH-T32-HL144470 to BTS).

Declaration of Interests

Drs Toll and Carpenter have served on an Advisory Board for Pfizer on e-cigarettes, and Dr Toll testifies on behalf of plaintiffs who have filed litigation against the tobacco industry.

Data Availability

The data that support the findings of this study are available from the corresponding author, AMR, upon reasonable request.