Motivational Interviewing to Encourage Quit Attempts Among Smokers Not Ready to Quit: A Trial-Based Economic Analysis

Abstract

Introduction

This study used data from a randomized controlled trial evaluating the efficacy of motivational interviewing (MI) relative to health education (HE) and brief advice (BA) to encourage quit attempts and cessation in order to determine their relative cost-effectiveness.

Aims and Methods

Urban community residents (n = 255) with low desire to quit smoking were randomized to MI, HE, or BA which differed in communication style and/or number of treatment sessions. Incremental cost-effectiveness ratios were used to compare the intensive interventions (MI and HE) to BA for facilitating quit attempts and smoking cessation. Costs were calculated from the perspective of an agency that might engage in program delivery. Sensitivity analysis examined different assumptions for MI training and pharmacotherapy costs.

Results

Total intervention delivery time costs per participant for MI, HE, and BA were $46.63, $42.87, and $2.4, respectively. Cost-effectiveness ratios per quit attempt at 24 weeks were $380 for MI, $272 for HE, and $209 for BA. The cost per additional quit attempt for MI and HE relative to BA was $508 and $301, respectively. The cost per additional quit for MI and HE relative to BA was $2030 and $752, respectively. Four separate sensitivity analyses conducted in our study did not change the conclusion the HE had a lower Incremental Cost-Effectiveness Ratio for both quit attempts and cessation.

Conclusions

HE was the most cost-effective of the three types of smoking cessation induction therapies and therefore may be preferable for smokers who are less motivated to quit. Providing valuable cost information in choosing different clinical methods for motivating smokers to quit.

Implications

All direct costs and activity-based time costs associated with delivering the intervention were analyzed from the perspective of an agency that may wish to replicate these strategies. A randomized controlled trial evaluating the efficacy of MI relative to HE and BA to encourage quit attempts and cessation determined their relative cost-effectiveness. HE was the most cost-effective of the three types of smoking cessation induction therapies and therefore may be preferable. Despite guideline recommendations, MI may not be the best approach to encourage quit attempts in diverse populations. Rather, a structured, intensive HE intervention might be the most cost-effective alternative.

Introduction

The Department of Health and Human Services Clinical Practice Guideline recommends the use of Motivational Interviewing (MI) for smokers who are currently unwilling to quit in order to encourage future quit attempts.¹ MI is a collaborative, goal-oriented style of communication designed to strengthen motivation and commitment to a specific behavior change goal.^2,3 Although several studies have indicated that MI can increase quit attempts and cessation among smokers,^4–6 significant methodological limitations have been noted, including the lack of comparison with intensity-matched controls, lack of evidence of fidelity in delivering MI, and failure to account for the level of initial motivation to quit.^4,6

The parent study described in this report, KC Quest, was designed to address these and other limitations.⁷ The study compared three alternative clinical methods for motivating smokers to quit: (1) Brief Advice (BA), (2) intensive Health Education (HE), and (3) intensive MI. While the trial was primarily designed to assess the efficacy of the alternative interventions, translation of the findings into practice also requires an evaluation of the relative cost-effectiveness of these interventions.

To date, cost-effectiveness studies of treatments for reducing tobacco use have focused primarily on interventions for smokers who are interested in quitting.^8–10 While these studies have shown that pharmacotherapy and counseling are highly cost-effective, the cost of treatment for smokers who are not motivated to quit is likely to be different from that of smokers enrolled in typical smoking cessation treatment trials. In trials where the goal is to encourage less motivated smokers to try to quit (referred to as “induction trials”), resources need to be devoted to strategies to motivate or encourage the unmotivated smokers to try to quit in addition to providing the usual resources to assist with any attempt to quit (ie, pharmacotherapy and counseling). In addition, the primary outcome of an induction intervention is whether or not an attempt to quit was made, whereas in a typical smoking cessation intervention the primary outcome is success in quitting (among a group of individuals motivated to try to quit). Theoretically, any effective induction intervention could be paired with any effective cessation treatment approach to help smokers, who have become motivated by the induction intervention, achieve abstinence. However, induction treatments may also vary in how well motivation is sustained through initial cessation and maintenance of cessation, which may also impact cessation rates. Induction treatments should thus be evaluated with respect to their effect on motivating attempts to quit as well as their effect on ultimate smoking cessation.

Relatively few studies have examined the cost-effectiveness of smoking cessation induction interventions such as MI which focuses on smokers who have not yet committed to try to quit, and very few of these have focused on MI or compared alternative behavioral interventions.^11,12 One prior study compared the cost of MI for smoking cessation with that of BA to quit among low-income pregnant women who were “either current smokers or recent quitters (eg, quit during previous 3 months).” ^13,14 Their results indicated that while MI was not cost-effective for smoking cessation, it was cost-effective for relapse prevention.¹³ Specifically, the cost-effectiveness of the MI intervention for relapse prevention was $851 per life year saved and $628 per quality-adjusted life year. Because pregnant women represent a unique treatment subpopulation, it is unclear whether these findings are generalizable to a broader population of adult smokers.^15–17 In addition, this study did not limit enrollment to participants with low motivation to quit (ie, the population for whom MI should be most cost-effective) and also did not examine the cost of achieving quit attempts.

In order to increase knowledge on the cost-effectiveness of MI interventions, the goal of this study was to determine the cost of treating unmotivated smokers with respect to achieving quit attempts and achieving cessation. We also considered the impact of including the related costs of conducting training and providing pharmacotherapy to participants who decided to quit.

Methods

Participants

The parent trial was a multi-arm, parallel, randomized trial conducted from November 2010 to November 2011 in a large Midwestern city.^7,18 We compared the use of MI, HE, and BA for smoking cessation induction among a diverse sample of adult smokers (n = 255) with low motivation to quit. Smokers were recruited and randomized to MI, HE, or BA, according to an imbalanced randomization allocation (2:2:1 for MI, HE, and BA, respectively). Detailed methods and results have been published elsewhere.¹⁸

Recruitment, Interventions, and Main Outcomes

Study participants were recruited through community-wide advertisements, physician referral, and word of mouth.⁷ Recruitment materials invited “smokers” or “smokers not quite ready to quit” to participate in a research study evaluating how health care providers should talk to smokers about their health. Of 766 individuals screened, 255 were randomized: 102 to MI, 102 to HE, and 51 to BA.¹⁸

Participants in the MI and HE intervention arms received four 24-minute sessions (two over the phone) over the course of 18 weeks, while the minimal intervention control condition, BA, involved one 5-minute session at baseline designed to mimic usual care. The primary outcome of the study was the occurrence of any self-reported quit attempt lasting at least 24 hours between randomization and the 24-week follow-up assessment, and the secondary outcome was biochemically verified 7-day point-prevalence abstinence at follow-up.⁷ Participants who decided to set a quit date within the study period were offered pharmacotherapy (varenicline or nicotine patch or lozenge) free of charge in order to be consistent with the standard of care for smoking cessation,¹ and those in the MI and HE arms received counseling in the formulation of a “quit plan” that incorporated proven behavioral strategies for smoking cessation.

Measures

Demographics and smoking characteristics were assessed at baseline (Table 1) and did not differ significantly between groups.¹⁸ Nicotine dependence was assessed using a single item¹⁹ from the Fagerström Test for Nicotine Dependence that assessed the time between awaking and smoking the first cigarette (0 = after 60 min, 1 = 31–60 min, 2 = 6–30 min, 3 = within 5 min).¹⁹ Use of any smoking cessation medications since the last assessment was measured at 12 and 24 weeks follow-up with a self-report checklist of various cessation pharmacotherapies. Motivation to quit smoking was assessed at baseline, by aggregating three self-reported items: (1) “motivation to quit” (0 = not at all, 10 = extremely), (2) “motivation to quit in the next 2 weeks” (0 = not at all, 10 = extremely), and (3) the Contemplation Ladder (0 = no thought of quitting, 10 = taking action to quit).^20,21 The primary outcome was the self-report of any serious attempt to quit smoking for at least 24 hours during the previous 3 months and was collected at baseline, week 12, and at the week 24 follow-up. The occurrence of any quit attempt between baseline and the week 24 follow-up period was calculated by collapsing any attempt across the 24 weeks. Additionally, self-reported 7-day point prevalence smoking abstinence was collected at week 12 and week 24, and abstinence was verified biochemically at week 24 using saliva cotinine.²²Table 1 summarizes outcome data from the original trial.¹⁸ Unexpectedly, HE yielded the highest percentage of quit attempts at week 24 (52.9%) though differences between HE and MI (45.1%) and BA (43.1%) did not reach statistical difference. Furthermore, HE yielded more verified cessation (7.8%) than MI (2.9%) or BA (0.0%),¹⁸ a difference that was statistically significant.

Table 1.

Baseline Demographic Information and Smoking Characteristics for the Participants

Characteristic	Motivational interviewing (n = 102)		Health education (n = 102)		Brief advice (n = 51)
			No.	(%)
Age in years	45.0	±11.7	46.7	±10.2	45.5	±10.5
Monthly income
Less than $1000	58	(57)	59	(58)	30	(59)
$1000–$2000	20	(20)	17	(17)	11	(22)
Greater than $2000	11	(11)	14	(14)	5	(10)
Declined to answer	13	(13)	12	(12)	5	(10)
Race/ethnicity
White, non-Hispanic	29	(28)	30	(29)	15	(29)
Black, non-Hispanic	68	(67)	68	(67)	31	(61)
Other, non-Hispanic	3	(3)	2	(2)	2	(4)
Hispanic	2	(2)	2	(2)	3	(6)
Smokes first cigarette within 5 min of waking up	48	(47)	49	(48)	31	(61)
At least one prior quit attempt	65	(64)	73	(72)	34	(68)
			Mean	±SD
Cigarettes per day	16.2	±9.0	16.9	±9.4	18.0	±10.8
Week 24 quit attempts (%)		45.1		52.9		43.1
Week 24 cessation (%)		2.9		7.8		0.0

No statistically significant differences between groups (all p > .05).

Economic Evaluation

For the intervention cost analysis, the direct costs and activity-based time costs associated with delivering the interventions were identified and analyzed (Table 2). All research-related costs that did not directly impact participants (such as planning the intervention, protocol-related trainings, research effort for the study investigators, and data analysis) were excluded from cost calculations. All direct costs and activity-based time costs associated with delivering the intervention were analyzed from the perspective of an agency that may wish to replicate these strategies and are consistent with recently reported economic evaluation studies.^23–30 All costs are reported in 2012 US dollars. Sensitivity analyses were conducted to show the effect of excluding/including MI training and, for the cessation outcome, the effect of excluding/including pharmacotherapy costs.

Table 2.

Direct Costs for Intervention

Item	Units	Cost per unit ($)	Total cost ($)	Cost per participant* ($)
Hardware (computers)	3	816.00	2448.00	9.60
Other computing	n/a	Variable	1149.19	4.51
Printing	n/a	Variable	243.72	0.96
Postage	n/a	Variable	156.25	0.61
Telephone
On-site telecommunications	n/a	Variable	1935.00	7.59
Cellular	n/a	Variable	2880.00	11.29
Total direct costs			8812.16	34.56

n/a = not available.

*n = 255.

Direct Costs Associated With the Intervention

The fixed direct costs included fixed costs related to telecommunications, printing, and postage (Table 2). Telecommunications fixed costs included three computers and monthly service charges for landline and cell phones for three study counselors. We estimated these time expenditures using a ratio later determined by the investigators and counselors involved in the study. Printing and postage costs included supplies used to print and mail appointment and educational materials. In analyses focused on the cessation outcome, the cost of pharmacotherapy (varenicline, nicotine patch, or nicotine lozenge), which was offered without charge to participants who set a quit date during the intervention period, was also estimated as a direct cost.

Activity-Based Time Costs Associated With the Delivery of the Intervention

Activity-based time costs included the cost of counselors for the total amount of time spent on intervention delivery (Table 4). Counselors provided four counseling sessions for MI and HE of approximately 24 minutes: two of these sessions occurred in-person at the study site and the other two sessions took place over the phone.

Table 4.

Intervention Delivery Time Costs by Treatment Method

Treatment technique	Minutes per session (mean)	n	Total time (min)a	Total cost ($)	Cost per participant ($)
Motivational interviewing
MI counseling session 1	25.6	102	2611.20	1253.38	12.29
MI counseling session 2 (phone)	21.7	95	2061.50	989.52	10.42
MI counseling session 3	26.2	96	2515.20	1207.30	12.58
MI counseling session 4 (phone)	23.4	89	2082.60	999.65	11.23
Total—MI	24.2 (mean)		9270.50	4449.84	43.63b
Health education
HE counseling session 1	29.7	101	2999.70	1439.86	14.26
HE counseling session 2 (phone)	22.8	97	2211.60	1061.57	10.94
HE counseling session 3	23.8	95	2261.00	1085.28	11.42
HE counseling session 4 (phone)	18.6	88	1636.80	785.66	8.93
Total—HE	23.73 (mean)		9109.10	4372.37	42.87b
Brief advice
BA session 1	5	51	255	122.4	2.40
Total			18 634.6	8944.61

MI = motivational interviewing; HE = health education; BA = brief advice.

The reduction in participants across sessions is a result of study attrition.

^aTime cost is estimated using published 2012 national median hourly wage for three health educators ($0.48/min per person), as described in Methods section.

^bTotal cost divided by n = 102.

Intervention delivery time costs per study participants are given in Table 4. Three counselors conducted the interventional counseling sessions.⁷ Counselors’ time was summed across the intervention sessions and valued at the median national wage plus a 25% fringe benefit rate for a health educator (occupation code 21–1091) from the US Bureau of Labor Statistics National Occupational Employment and Wage Estimates.³¹ Using this government-published 2012 data, it was determined that the national hourly median wage for a health educator was $23.04. Inclusion of the 25% fringe benefit rate increased this hourly value to $28.80, which equates to a cost of $0.48 per minute of the counselors’ time. In other research or clinical settings, the salaries can be altered to correspond to the type of employees assigned to implement the selected intervention.

Protocol Specific and Initial Training Time Costs

All three of the master’s-level counselors had previously received initial MI training for prior research studies. Therefore, for the purpose of the current intervention delivery, the counselors received “booster” training specific to the MI, HE, and BA protocols for smoking cessation in this trial (Table 3). We refer to this as “protocol-specific training” and this should be distinguished from the more extensive initial training and practice that is required to learn to deliver MI in general. This is a key distinction between MI and the other treatments, as HE and BA do not require any comparable initial training for master’s-level professionals. Protocol-specific training for MI included a half-day training workshop, video demonstrations, reading materials, instructions specific to the study protocol, and role-play practice sessions with feedback from supervisors in a group setting (Table 3). The protocol-specific expert-led training costs were not included in main costs analyses (basic scenario); however, they were used in the sensitivity analyses so that costs of the intervention can be considered both with and without the costs of MI training specific to the intervention. For initial general MI training cost, we included estimated cost of expert-led MI training using past findings.^2,32 In these articles, training costs were presented in a variety of ways for decision makers.^2,32 Representing expert-led training costs and strategies needed for teaching clinicians about MI. These standards of MI performance have been commonly used to certify clinicians in MI effectiveness trials. Total economic cost of expert-led MI training was $5638 (in 2006 dollars)^2,32 which was used for the purpose of this sensitivity analysis as an alternative scenario relating training of initial professionals.^2,32

Table 3.

Protocol-Specific Training Costs

	Total time (min)	Total cost ($)	Time (min) per participant	Cost ($) per participant*
MI training (80 h) for three counselors	14 400	6912.00	141.18	67.76
HE training (28.5 h) for three counselors	5130	2462.40	50.29	24.14
BA training (3 h) for three counselors	540	259.20	10.59	5.08
Total initial time costs	20 070	9633.60	—	—

MI = motivational interviewing; HE = health education; BA = brief advice.

*Total n = 255, MI n = 102, HE n = 102, BA n = 51, only included in sensitivity analysis.

Cost-Effectiveness Analysis

We performed a cost-effectiveness analysis (CEA) to compare the relative cost of BA, HE, and MI. Effectiveness was determined with respect to both quit attempts and smoking cessation. The CEA methodologies used were as follows: (1) ICERs, defined as the ratio of the difference in costs to the difference in effectiveness between two competing treatment strategies^23,33,34 and (2) ACERs, which estimate average cost per effect. Here effectiveness is calculated by indicating the percentage of quit attempt or cessation in one arm (ie, MI, HE, or BA) relative to another arm (Table 1).

To assess the cost-effectiveness of quit attempts, we created decision tree models³⁵ at week 24 when the final quit attempt data were collected (Appendix, Figure 1). The expected quit attempts is the average number of quit attempts weighted by probability of successful or failed quit attempt event in the decision tree.³⁵ Similar analyses were not feasible for the cessation outcome due to the low numbers of individuals who had quit at follow-up. However, we examined ACERs and ICERs for MI and HE using the 24-week cessation data as given in Table 5. BA had to be excluded from ACER and ICER because no one in this intervention group reported cession at week 24.

Table 5.

Overall Intervention Delivery Cost by Treatment Method and Cost-effectiveness Analysis*

Cost-effectiveness analysis
Outcomes	Quit attempt at 24 weeks				Cessation/quit~
	n a	ACER		ICER	n	ACER		ICER
MI	21	$380		$508	3	$2658		$2030
HE	29	$272		$301	8	$987		$752
BA	9	$209		n/a	0	n/a		n/a
Cost-effectiveness analysis of quit attempt with uncertainty
Intervention	Cost($)	Quit attempts	ACER	90% UIb-ACER (low)	90% UIb-ACER	ICER	90% UIb-ICER (low)	90% UIb-ICER (high)
BA	2487	17	146	94	222	n/a	n/a	n/a
HE	11 309	40	283	163	482	472	171	917
MI	15 998	35	457	302	685	890	413	1546

*Excluded protocol-specific training costs.

~Included pharmacotherapy costs.

^aThe expected quit attempts is the average number of quit attempts weighted by probability of successful or failed quit attempt event (TreeAge). ACER = average cost-effectiveness ratio; ICER = incremental cost-effectiveness ratio; MI = motivational interviewing; HE = health education; BA = brief advice.

^b90% uncertainty interval (UI).

At 24 weeks, Catley et al. found that the quit attempt percentages are 45.1% for MI, 52.9% for HE, and 43.1% for BA.¹⁸ This gives 54 quit attempts for HE (n = 102), 46 for MI (n = 102), and 22 for BA (n = 51). These were the inputs that fed into the TreeAge model. Quit attempts 21 for MI, 29 for HE, and 9 for BA are the output of the model. These are predicted quit attempts calculated as a sum of all possible values—each multiplied by the likelihood of occurrence (in our case—quit attempt or no quit attempt). Probability sensitivity analysis was performed and distributions were used to describe quit attempts (derived from trends and data from original study) and costs (derived from variability in costs given in different scenario in Table 6). These distributions were used as inputs to the uncertainty simulation model.

Table 6.

Sensitivity Analysis for Overall Intervention Costs and Cost-Effectiveness Analysis for Quit Attempt and Cessation Outcomes

	Costs according to scenarios ($)		Cost-effectiveness change from basic scenario ($)
	Total cost	Per participant	ACER*	ACER~	ICER*	ICER~
Scenario 1: Vareniclinea
MI	$27 474.70	$269.36	$9158	$1308.32	$7230	$1807
HE	$32 857.23	$322.13	$4107	$1133.01	$3384	1353.62
BA	$5784.83	$113.43	n/a	$642.76	n/a	n/a
Scenario 2: nicotine replacement therapyb
MI	$12 774.70	$125.24	$4258	$608.32	$3310	827.49
HE	$14 041.23	$137.66	$1755	$484.18	$1400	$1170.10
BA	$2844.83	$55.78	n/a	$316.09	n/a	n/a
Scenario 3: MI trainingc
MI	$13 612.70	$133.46	$4537.57	$648.22	$3909.29	$977.32
HE	$7897.23	$77.42	$987.15	$272.32	$751.55	$300.62
BA	$1884.83	$36.96	n/a	$209.43	n/a	n/a
Scenario 4: Protocol-specific trainingd
MI	$14 886.70	$145.95	$4962.23	$708.89	$4247.56	$1061.89
HE	$10 359.60	$101.56	$1294.95	$357.23	$1026.95	$410.78
BA	$2144.03	$42.04	n/a	$238.23	n/a	n/a

ACER = average cost-effectiveness ratio; ICER = incremental cost-effectiveness ratio.

*For quit attempt; ~for cessation outcomes.

^aIncludes estimated cost of a 12-week supply of varenicline for 62 participants who chose to use pharmacotherapy in the trial (BA = 5; HE = 32; MI = 25) (16, 34).

^bIncludes estimated cost of a 12-week supply of nicotine replacement patches for 62 participants (BA = 5; HE = 32; MI = 25) (16, 35).

^cIncludes estimated cost of expert-led MI training using past findings (2, 30). Total expert-led MI training cost was $5638 (2, 30).

^dIncludes cost of protocol-specific training for MI, HE, and BA techniques (reported in Table 3).

Sensitivity Analysis

The sensitivity analysis estimated the robustness of the economic evaluation.^23,26 In this trial-based economic evaluation, we considered alternative scenarios for both quit attempts and cessation outcomes gathered from our parent randomized trial.^7,18 We report average cost-effectiveness ratios (ACER) and incremental cost-effectiveness ratios (ICER) related to four different scenarios that may be relevant for other research or clinical settings (Table 6).

Scenarios 1 and 2 included the estimated cost of two different pharmacotherapy options. Because pharmacotherapy was only offered after a participant set a quit date, it was considered not relevant to whether or not a quit attempt was made. Scenarios 1 and 2 therefore only examined the cost of pharmacotherapy for the cessation outcome. Scenario 1 considered the cost of varenicline which was encouraged during the trial. Although the exact cost of the drug varies, published estimates average $780 (cost after a $75 manufacturer coupon) for a 12-week supply.³⁶ Scenario 1 estimates the range of costs for supplying varenicline to 62 participants (BA = 5; HE = 32; MI = 25) for 12 weeks, corresponding with the participants who received the drug in this study. Scenario 2 considered the cost of nicotine replacement therapy which can be offered as an alternative to varenicline; the costs of a 12-week supply of nicotine replacement patches ($192 for six boxes each containing a 2-week supply for approximately $32 each)³⁷ are included in Scenario 2. Given results of prior research examining the effectiveness of different pharmacotherapy types, results related to cessation outcomes assume equivalent effectiveness of varenicline and nicotine replacement therapy.¹¹

Scenarios 3 and 4 considered the additional cost of providing initial MI training to counselors and examined the quit attempt and cessation outcomes.^7,18 Scenario 3 included the costs of training counselors in general MI principles and methods. This type of training was only used to deliver the MI intervention. Because our counselors were already trained in MI, we estimated the additional cost of 15 hours of expert-led MI training based on a published report of MI training costs.³² For the sake of consistency across costs, the values reported from the report were converted from 2006 US dollars to 2012 dollars, using a 13.9% cumulative inflation rate.³⁸ Total expert-led MI training total cost was $5638.^2,32 In Scenario 4 we included the costs of protocol-specific training in smoking cessation for MI, HE, and BA (taken from Table 2).

To further explore the robustness of the CEA, we performed a probabilistic sensitivity analysis (PSA) that considered the uncertainty in the cost and number of quit attempts per intervention arm. Probability distributions were used to model the variability in these parameters. Costs were assumed to be gamma distributed (BA: α = 0.50, β = 1461, shift = 1754; MI: α = 0.53, β = 7960.5, shift = 11763; HE: α = 0.49, β = 7960.5, shift = 7405). We arrived at these distributions by observing the cost variations in the different scenarios given in sensitivity analysis (Table 6). These variations were captured in cost functions that led to generation of costs data that were well represented by gamma distributions. Quit attempts were described by beta distributions (BA: α1 = 0.91, α2 = 0.91, a = 13.3, b = 21.5; MI: α1 = 0.91, α2 = 0.91, a = 25.2, b = 44.5; HE: α1 = 0.91, α2 = 0.91, a = 26, b = 54).^39,40 Beta distributions best described the variations we observed in quit attempts. In the work of Catley et al.,¹⁸ we observed that for 0, 3, and 6 months, motivation to quit was correlated with quit attempts. A linear function relating time in months and quit attempts was derived and used to generate enough quit attempt data that were best described by beta distributions. The number of samples drawn was set at 2000 to allow cost estimates to stabilize. A 0% discount rate was included in quit attempts and a 6% discount rate was included for cost.⁴⁰ A 90% uncertainty interval around ACER and ICER was reported as a measure of uncertainty. These measures were chosen based on the World Health Organization guidelines for cost-effectiveness.⁴⁰ The 90% uncertainty interval was generated using the simple percentile method, which involves omitting the lower and upper 5% of estimates.^18,40

TreeAge and @RISK 7.0 (Palisade Decision Tools) were used to tabulate the scenario-specific ACER and ICER of quit attempts, as well as calculated overall and patient-specific time and costs (Table 6).³⁵

Results

Baseline Demographic and Smoking Data

Baseline demographic information and smoking characteristics for the participants are given in Table 1.

Cost Analysis

Direct Costs

Total direct costs, which are summarized in Table 2, equaled $8812.16 ($34.65 per participant). The direct cost included three computers ($2448), printing ($243.72), postage ($156.25), telephone services ($4815.), and additional computing charges (eg, E-mail accounts and miscellaneous departmental expenses) ($1149.19). Printing costs ($0.96 per participant) included supplies used to print appointment reminder cards, educational resources, and counseling scripts. Postage costs ($0.61 per participant) included stationery and postage to mail appointment reminders and letters. Telephone-based costs were accumulated over the intervention delivery period. Telecommunication costs ($18.89 per participant) included the monthly service fee for three landline phones ($7.59 per participant) and three cellular phones ($11.29 per participant).

Protocol-Specific Training Time Costs

The protocol-specific time cost (Table 3) accounts for training time for three counselors to implement MI, HE, or BA for smoking cessation induction and cessation. These training costs are included only in the sensitivity analysis (Scenario 4) as training is a one-time expenditure per counselor and would not be applicable in clinics employing staff already trained in protocols for smoking induction and cessation.

Activity-Based Time Costs Associated With the Delivery of the Intervention

The total cost per individual participant by treatment arm is provided in Table 4. Only activity-based time costs associated with intervention delivery are included in this analysis. The total time cost of intervention delivery ($8944.6) represents an aggregate of the overall delivery time costs for four MI sessions ($4449.84), four HE sessions ($4372.37), and one BA session ($122.4). MI and HE sessions were designed to be of similar duration and were completed at similar rates (four 24-min sessions), resulting in similar total and per participant time cost for MI and HE. BA, which consisted of one brief session, was around one-fifth of the cost of MI and HE.

Overall Cost and CEA

The total intervention delivery time costs per participant for MI, HE, ad BA were $43.63, $43.87, and $2.40, respectively (Table 4). Table 5 gives the cost-effectiveness of each of the counseling methods based on these costs for each outcome. For quit attempts the decision tree model (Appendix, Figure 1) revealed MI to have the highest ACER per quit attempt at 24 weeks and BA the lowest (Table 5). The ICER for MI relative to BA was also higher than for HE relative to BA. For the cessation outcome, results similarly revealed that MI had a higher ACER than HE (Table 5). The ACER for BA could not be calculated as no one in this treatment arm achieved cessation. A decision tree model was not constructed due to the small sample size. The ICER for MI relative to BA was also much higher than for HE relative to BA.

Sensitivity Analysis of Overall Intervention Costs and Cost-effectiveness

Sensitivity analyses of overall intervention costs showed how costs changed in each scenario with the effect of including other related intervention costs (Table 6). Change in cost-effectiveness ratios was reported using ACER and ICER using trial outcomes (quit attempts and cessation). In Scenario 1, including the costs of the two different medications substantially increased all interventions costs for the smoking cessation outcome, but MI still had a higher ACER and ICER than HE. Similarly, including either general MI training or specific protocol training for each of the interventions substantially increased intervention costs but increased MI costs the most so that the ACER and ICER remained higher for MI for both quit attempt and cessation outcomes. Scenarios of sensitivity analyses represented how each sensitivity assumption (scenario) changed costs compared with each other and the resulting comparative costs and effectiveness for each treatment group.

The PSA of cost-effectiveness showed a similar outcome with HE having the lowest ACER (compared to MI) and ICER as the preferred intervention.⁴¹ In Table 5 we have given that, both in the point estimate of ACER and ICER, HE is low compared to MI. The 90% uncertainty interval for HE overlaps with that of BA and MI. It is worth mentioning that BA has a lower ACER compared to MI, however its 90% uncertainty interval overlaps with that for of HE (Table 5). The interpretation is that it is not possible to be 100% sure that BA would be the best intervention, although at low levels of resource availability (eg, at a total cost of $2487), BA is the only intervention because the other interventions are not affordable. In a separate analysis after adding 2000 simulations, we observed that 82% of the times HE was the better intervention strategy (not given in tables).^12,41

Discussion

The present study shows the value of alternative smoking induction interventions for smokers with low motivation to quit. The previously reported efficacy outcomes unexpectedly showed that HE was equally, if not more, effective than MI for quit attempts and cessation. Consistent with those findings, the cost-effectiveness analyses favored HE over MI. Results revealed that HE had the best cost-effectiveness ratio per quit attempt ($380, $272, and $209 for MI, HE, and BA, respectively) and per verified quit ($2658 and $987 for MI and HE, respectively). Incremental cost-effectiveness for quit attempts and cessation outcomes was substantially lower for HE than for MI relative to BA ($301 vs. $508 and $752 vs. $2030 for quit attempts and cessation, respectively). In our study we found that $752 (HE) to $2030 (MI) were needed to achieve additional quit outcomes among smokers who are not ready to quit. PSA also favored HE but there was overlap in the 90% uncertainty interval, indicating that our results do not preclude the possibility of equivalence between cost-effectiveness of arms.

Because there are few efficacy trials of smoking cessation induction interventions and even fewer economic evaluations of MI, our study is best compared with a prior study that compared a single motivational intervention to BA among 536 smokers during a single visit with their general practice physician in the United Kingdom.⁴² In that study the incremental cost per quit attempt including/excluding training costs was £311.99 (US$504.76)/£183.47 (US$296.83), values that are closer to the incremental cost-effectiveness of MI in our study (including/excluding protocol-specific training costs $974/$383). There are notable differences between the prior study and the present study including the use of one (slightly longer) session compared to the use of four sessions in our study. The reasons for the difference can also be due to differences in setting (UK primary care practice vs. US university laboratory), smoker demographics, intervention providers, and their practitioner’s relationship with patient.^43,44

Our findings can also be compared to studies that have examined cost-effectiveness of other cessation induction interventions. Cromwell et al.³³ evaluated the cost-effectiveness of implementing the Clinical Practice Guideline which involves screening, motivating, and assisting all smokers at every primary care visit to quit. Direct comparison of costs with that study is not feasible because of substantial differences in methodology. For example, the authors assumed screening and advising would need to be done by primary care physicians before handing off to less expensive providers. They therefore included the high cost of the time spent by the physician as well as the overhead related to maintaining a physician practice.

One other induction study involved a mass media antismoking campaign in England which was based on behavioral change therapy (motivational text messaging and an app with ongoing support and self-monitoring tools). The intervention was found to have an ICER of approximately $704.93 (95% confidence interval: $159.25 to $1250.61).⁴⁵ Although direct comparison with this study is also complicated by the use of different methods (ie, a published standard model for calculating the cost-effectiveness of smoking interventions), contexts (ie, different countries), and types of intervention (eg, clinical vs. mass media), results do suggest that the costs of different behavioral interventions do not diverge that widely.

A recent meta-analysis of smoking cessation intervention studies for smokers not ready to quit used estimated costs to assess the cost-effectiveness of behavioral interventions including MI.^12,30 The costs and cost-effectiveness estimated in this meta-analysis are substantially higher than in ours. The main reason appears to be differences in how counseling session costs were calculated. Whereas ours are based on labor costs for the duration of the session, the meta-analysis based costs on prior studies that included additional costs such as a pro-rata share of time spent on activities other than direct patient care, the cost of clinic space, time involved in scheduling sessions, reminding patients of visits, and recording sessions in medical records.

An important contribution of the present study is the comparison between alternative behavioral interventions. Our results indicate that MI has a lower cost-effectiveness due to the modest effect on quit attempts, rather than a higher cost as there were minor differences in time and costs between MI and HE. This is consistent with relatively modest effects of MI on smoking cessation, compared to other addictions that have been previously noted.⁴⁶ The advantage of HE was increased when we included the counselor training costs. MI was particularly costly with respect to training because counselors need fairly extensive training in the general principles and methods, including participation in a clinical workshop, plus individual coaching sessions and practice feedback, to be effective.⁴⁷ A review of 10 MI training studies⁴⁰ determined that the average duration of MI workshops was 9 hours, and maintenance of MI skills was best sustained when follow-up training sessions were repeated over the long term. These extensive communication skills are not needed to deliver semi-scripted HE. Training costs to deliver the specific smoking cessation protocols of this study also differed somewhat with MI being the most costly.

The advantage of HE was also not affected by considering medication costs. Increased medication costs are due to the increased number of quit attempts produced by HE. Although medication costs per participant were higher for HE than for the other groups, those costs were outweighed by the impact on quitting. HE had a substantially lower ACER and ICER than MI when accounting for medication, though both were more cost-effective than providing BA, which did not impact quitting. Based on previous studies indicating equal effectiveness of varenicline and nicotine replacement therapy,¹¹ nicotine replacement reduces the cost of treatment because it is less expensive. Taken together, the results point to a significant cost-effectiveness advantage of HE over MI across all scenarios.

When considering the cost-effectiveness of HE, HE was not only superior to MI but was also more cost-effective than BA, even though BA was far less time-intensive. This is important because it suggests that the additional resources required to implement an intensive HE intervention to encourage smokers to quit (relative to providing BA) is warranted. The failure of any individual in BA to achieve cessation also highlights the necessity of assisting smokers with their efforts to quit once a motivational intervention is successful at encouraging a quit attempt. Furthermore, both MI and HE provide long-term benefits from smoking cessation on patient’s well-being, contributing to a reduction in health care expenses that likely outweigh the costs of both of these interventions.⁴⁰

Limitation

It should be noted that indirect costs incurred by participants in these interventions were not included. These costs may include travel time, mileage, cellular phone usage time, and lost wages due to taking time off from work. While these are important costs to bear in mind for future studies, the retrospective nature of this analysis prevented us from obtaining this information. Prospectively tracking specific information such as the nature and duration of phone calls (ie, recruitment vs. scheduling and intervention delivery) could have strengthened this economic evaluation. Nevertheless, the great majority of the cost of these interventions relates to counselor time cost which was prospectively tracked. Another limitation is that, as noted above, we did not calculate any future savings, such as nonmedical attributable savings or costs associated with early death from smoking. These values are certainly worth examining and should be used in future investigations. Caution is also warranted in interpreting differences between the arms in this study given the PSA results, and results should not be generalized to populations of smokers that differ from the predominantly low-income sample of African American smokers included in this study. Lastly, as noted above, the lack of standardization in economic evaluation makes it difficult to compare across studies.³⁹ Standardization of economic evaluations with transparency and rigorous analysis would help to make future studies more comparable particularly when interventions and programs vary greatly in cost-effectiveness depending on contextual factors.⁴⁰

In conclusion, this is the first economic analysis of an MI-based smoking cessation intervention aimed at smokers with low motivation to quit. Randomized controlled trials are considered to be the gold standard of clinical evidence, therefore conducting economic evaluations like our study alongside controlled trials enhances the reliability and validity of findings.⁴⁰ Our analysis provides valuable cost information for agencies interested in choosing among different clinical methods for motivating smokers to quit.

Funding

This study was supported by a grant (R01 CA133068) from National Institutes of Health (NIH), National Cancer Institute (PI, DC). Pfizer provided varenicline (Chantix) through Investigator-Initiated Research Support (No. WS759405). This study “A Trail-Based Economic Evaluation of Clinical Interventions to Motivate Smokers to Quit” was also funded by Frontier Trail Blazer Award (PI, RR), The Heartland Institute for Clinical and Translational Research. This project was supported by an Institutional Clinical and Translational Science Award, NIH/National Center for Advancing Translational Science grant number 8UL1TR000001-02.

Declaration of Interest

None of the authors reports any conflicts of interest. Catley and Goggin occasionally conduct Motivational Interviewing training.