Cost-Effectiveness of Smoking Cessation Approaches in an Emergency Departments

Abstract

Introduction:

Americans of lower socioeconomic status use tobacco products at disproportionately high rates and are overrepresented as patients of Emergency Departments (EDs). Accordingly, ED visits are an ideal time to initiate tobacco treatment and aftercare for this vulnerable and understudied population. This research estimates the costs per quit of ED smoking cessation interventions and compares them with other approaches.

Methods:

Previously published research described the effectiveness of two multicomponent smoking cessation interventions including brief negotiated interviewing (BNI), nicotine replacement treatment (NRT), quitline referral, and follow-up communication. Study 1 (collected 2010–2012) only analyzed the combined interventions. Study 2 (collected 2017–2019) analyzed the intervention components independently. Costs per participant and per quit were estimated separately, under distinct “intervention with dedicated staff’ and “intervention with repurposed staff’ assumptions. The distinction concerns whether the intervention used dedicated staff for delivery or whether time from existing staff was repurposed for intervention if available.

Results:

Data were analyzed in 2021–2022. In the first study, the cost per participant was $860 (2018 dollars) and the cost per quit was $11,814 (95% CI: $7,641 - $25,423) (dedicated), and $227 per participant and $3,121 per quit (95% CI: $1,910 - $7,012) (repurposed). In Study 2, the combined effect of BNI, NRT, and quitline cost $808 per participant and $6,100 per quit (dedicated) (95% CI: $4,043 - $12,274), and $221 per participant and $1,669 per quit (95% CI: ($1,052, $3,531) (repurposed).

Conclusions:

Costs varied considerably per method used but were comparable to other smoking cessation interventions.

Introduction

Smoking remains the leading cause of preventable death in the United States, with about 480,000 deaths per year.^1,2 Over the past 50 years, smoking prevalence has dropped from 42% to approximately 12%, saving 8 million lives.³ However, the decline in smoking and tobacco use has been slower among individuals of lower socioeconomic status,⁴ so strong incentives suggest targeting interventions for this population.

Emergency Departments (EDs) disproportionately treat individuals with lower incomes and irregular access to primary care.⁵ Hence, ED visits present an opportune time to discuss patients’ tobacco use and initiate tobacco treatment and aftercare⁶ for this vulnerable and understudied population.

Relatively little past research has demonstrated the costs associated with implementing smoking cessation interventions in EDs.^7,8 Herein are estimates of the costs associated with two successful smoking cessation interventions among adult ED patients.^9,10 Both intervention studies were similar in that the interventions examined included four components: counseling, nicotine replacement therapy (NRT), quitline referral, and follow-up communication. However, the design used in the 2021 study allowed the impact and costs associated with each of the four components of the intervention to be evaluated separately.

Methods

Study Sample

The study involved economic analysis of data collected from two smoking cessation studies. In Study 1,⁹ 390 adults who smoke were recruited in the ED and block randomized to receive a multicomponent smoking cessation intervention. Another 390 adults who smoke recruited from the ED were randomly assigned to serve as controls. However, two participants originally allocated to the intervention arm were duplicate registries, and two were deceased by the 3-month follow-up. Using an intent-to-treat approach, costs were based on the 388 participants who received the intervention (although only 386 completed the follow-up). For the control sample, 390 were successfully allocated, but two were deceased by the time of the 3-month follow-up. Study recruitment lasted 20 months.

In Study 2,¹⁰ among the study’s 1,056 participants, 528 were allocated to receive each intervention component and 528 were allocated to not receive that component. The experimental design enabled estimation of the separate effect of each intervention component. Only BNI and NRT were associated with statistically significant changes in smoking quits at 3 months.

Measures

Measures were effectiveness outcomes reported in two smoking cessation studies^9,10 along with the costs conduct the study interventions.

Statistical Analysis

Our initial review of the literature on the cost effectiveness of smoking cessation programs indicated variability in how costs were estimated. To support comparisons with the literature, two distinct, generalized approaches were used to estimate costs.

The first and sounder approach (described hereafter as Method A) follows an “intervention with dedicated staff’ philosophy based on labor costs as the percentage of full-time equivalents (FTEs) allocated to program delivery. This assumes dedicated intervention staff and is more broadly comprehensive. The first practical consequence of this approach is that costs include “idle time” for staff who recruit people and deliver the cessation treatment. For example, even if it requires only 15 minutes to deliver an intervention component to a single participant, if the program staff had to spend 5 hours waiting for an eligible patient to arrive in the ED, handling paperwork, etc., those ancillary costs are included. Second, the costs include funding for supervision, i.e., people managing intervention staff clinically and administratively. Third, this estimate includes costs incurred by the state for delivering medicine and counseling through its quitline. Thus, the Method A cost per participant and cost per quit reflect what funds would be required to implement an ED intervention with dedicated staff who cannot routinely be diverted to other tasks, thereby missing intervention opportunities.

Our second approach (described hereafter as Method B) is the most frequent approach seen in the literature. It follows an “intervention with repurposed staff’ approach and mirrors the cost that an ED or health care organization might incur delivering the intervention using existing staff, repurposing their time as needed if they are available. Method B assumes that when staff were not actively working on intervention-related activities, their costs were covered elsewhere. Following Method B, whenever possible, labor costs were based on the mean time spent to deliver intervention components (multiplied by loaded labor rates, which include fringe benefits and institutional indirect costs). When this information was not available, minutes from other data were estimated (described below). Unwisely, the literature does not count supervisory costs as costs of the intervention; rather, it assumes that responsibility is supported through existing supervisory structures. Unlike some published studies, this study included costs incurred by the state for delivering quitline medicine and counseling in Method B costs.

Each of the two studies^9,10 we examined included four components: a brief negotiated interview (BNI; a brief intervention based on motivational interviewing designed for medical settings) by a trained research associate; provision of a 6-week supply of nicotine replacement therapy (NRT—6 weeks of nicotine patches and gum); an active referral to the state’s quitline; and follow-up communication (booster calls in Study 1, and 6 weeks of SmokefreeTXT in Study 2).¹¹ Both the intervention and control groups received an informational brochure which included information about the quitline. The primary outcome in both studies was the proportion of subjects who were confirmed abstinent at 3 months. Subjects who self-reported past 7-day abstinence were asked to return to the ED to measure exhaled breath carbon monoxide with the Bedfont Micro+ breath carbon monoxide monitor. Both studies were approved by the Human Investigation Committee of Yale University. All participants provided written informed consent.

While experimental participants in 2015 received a single intervention with all four components, the 2021 study employed a Multiphase Optimization Strategy (MOST) design¹² in which a randomized 2⁴ factorial optimization trial was conducted. Across the 16 experimental conditions, half of the participants were assigned to receive or not receive each component. All possible combinations of components were represented, ranging from no components to all components. This allocation approach allowed the unique effect of each component to be estimated against the absence of that component.

BNI.

Under Method A, since staff only have intervention responsibilities, the cost of the BNI condition was based on the FTE percentage of time dedicated to recruitment and delivery of the intervention multiplied by loaded staff rates including salary, fringe, and overhead, plus the cost of supervision. Under Method B, the cost was based on loaded rates times the actual minutes devoted to intervention activity, with downtime assumed to all be repurposed.

NRT.

Using Method A, the cost of the NRT condition was based on the percentage FTE spent on NRT delivery and recruitment, plus supervision and the cost of the medication. For Method B, the research did not record the minutes required to deliver the NRT portion of the intervention. Instead, the cost of the BNI intervention was multiplied by the ratio of the FTE spent on NRT to that spent on BNI, in addition to the cost of the medicine.

Quitline.

Under Method A, quitline costs included staff time to recruit subjects and refer them to the Connecticut Quitline, plus supervision, as well costs to the state for registration, administration of counseling, and medicine through the quitline program. Under Method B, quitline costs were estimated by multiplying the Method B cost of the BNI intervention by the ratio of the FTE spent on NRT to the FTE spent on BNI, then added the costs the state incurred for quitline services.

Communication.

Both Bernstein et al.⁹ and Bernstein et al.¹⁰ used follow-up communication. In Study 1, Method A, the costs of booster telephone calls were based on the loaded labor rates and time for a trained coach to make calls to intervention participants. Under Method B, they were based on actual minutes to attempt to reach participants and complete the call. In Study 2, SmokefreeTXT text messages were delivered via a texting platform developed and managed by a subcontractor. The cost for both methods is based on the cost of that subcontract with Method A adding 10% FTE for recruiting participants and registering them with the texting service.

We reported all costs in 2018 dollars, using the Cost Price Index – All Items as an inflator when required. In Study 1,⁹ costs for a single intervention (all components) combined were compared with treatment as usual. The supervision and recruitment costs could not be disaggregated by intervention component. With the factorial design in Study 2,¹⁰ the cost of each intervention component, plus costs for combinations of components, was computed. Recruitment and supervision costs were not aggregated when components were combined. For comparability, costs for both studies used the staff wage, fringe, and indirect rates from Study 2.

We calculated costs from a societal perspective. Because the interventions were delivered during waiting periods at the hospital, no value was placed on the time spent by those being counseled. Their opportunity cost while waiting in a hospital treatment room or holding area is minimal. As well as producing average cost per quit, the incremental cost per quit as components operated in tandem were considered. Oracle’s Crystal Ball add-in to Excel with 100,000 iterations was used to bootstrap standard errors of the intervention costs and costs per quit, as well as the probability that adding an intervention would increase quits.

Results

The ED had dedicated smoking cessation staff for 12-14 hours daily during study recruitment periods of 20 months for Study 1 and 27 months for Study 2. All staff had at least a bachelor’s degree. Study investigators trained them in motivational interviewing and tobacco dependence treatment. Based on a 37.5-hour week, cessation staff had average hourly rates of $28.25 (in 2018 dollars), or $61.28 when loaded. The supervisor’s hourly rate was $74.32, or $161.22 when loaded.

For Costs per Intervention, in all study conditions, cessation staff spent 30% FTE annually on recruitment over a multiyear intervention period. If a patient received two or more interventions, that cost was only incurred once.

Regarding BNI using Method A. Spread across multiple people, cessation staff spent 30% FTE on BNI delivery. A supervisor spent 5% FTE annually. For Method B, excluding recruitment and idle time between participants, service delivery staff spent an average of 13.68 minutes per patient (standard deviation, 3.74 minutes) delivering the BNI intervention.

Regarding NRT using Method A, cessation staff spent 40% FTE annually on NRT delivery plus 30% FTE on recruitment. A supervisor spent 5% FTE annually. The 6 weeks of NRT medicine per person served cost $170. For Method B, since NRT delivery used one third more FTEs than BNI (40% FTE / 30% FTE), actual NRT delivery labor was estimated at an average of 18.24 minutes (13.68 * 4/3). Like in Method A, the cost of 6 weeks of NRT medicine was added.

Under Method A, the brochure given to control participants included information about the state’s quitline program, so both intervention and control participants used quitline services to some degree. The intervention costs included 10% FTE annually for cessation staff to register intervention participants with the quitline, but no control costs. Providing the brochure added no net cost, as both intervention and control participants received it. The quitline vendor provided information that the average cost for the quitline was $110 per user—$90 of which supported counseling and $20 NRT. This counseling and NRT were independent of the BNI counseling and NRT provided directly from the intervention. In Study 1, 124 intervention participants and 73 control participants received quitline counseling (net cost = (124 – 73) × $90 = $4,590), while 75 intervention subjects and 68 control subjects received NRT through the quitline (net cost = (75-68) × $20 = $140), for an average cost of $12 per person referred to the quitline. In Study 2, 62 intervention participants but no control participants received quitline counseling (net cost = 62 × $90 = $5,580), while 85 intervention subjects and 2 control subjects received NRT through the quitline (net cost = (85-2) × $20 = $1,660), for an average quitline cost of $14.

For Method B, since quitline referral used only one third of the FTEs used for BNI (10% FTE / 30% FTE), quitline delivery was estimated at an average of 4.56 minutes. The incremental costs of quitline delivery (registration, counseling, and medicine) were added.

Under Method A, cessation staff devoted 10% FTE to enroll staff in the intervention. In Study 1, making booster calls to participants added another 10% FTE. In Study 2, delivering the SmokefreeTXT text messages cost $66.84 per person.

Like with quitline referral, under Method B, an estimated 4.56 minutes were spent enrolling each participant. In Study 1, cessation staff made booster calls averaging 4 minutes to 285 intervention participants. Multiple efforts to reach the remaining 103 participants proved fruitless with 1 minute spent per person not reached. Thus, service delivery staff spent 1,243 minutes on booster calls, an average of 3.20 minutes per participant. In Study 2, texting at $66.84 per person substituted for these calls.

The costs and cost-effectiveness of the study interventions were estimates. Table 1 shows intervention effectiveness, cost per participant and per quit based on dedicated, supervised staff. In Study 1, the cost of the four-component intervention was the sum of the 30% FTE recruitment cost plus the non-recruitment costs of the four intervention arms. It totaled $860 per participant. Table 2 shows equivalent information with as-available staff and no dedicated supervision. The cost per participant in Study 1 totaled $227 including $45 for labor and $182 for medicine and quitline services. With the study’s 7.3% quit-rate above treatment-as-usual,⁹ the cost per quit was $11,814 with dedicated staff (95% CI: $7,641 - $25,423) and $3,112 with repurposed available staff (95% CI: $1,910 - $7,012).

Table 1.

Cost per participant, intervention effectiveness, and cost per quit based on dedicated, supervised staff.

Treatment	Cost/Participant (95% CI)	% Reduced (95% CI)	Cost/Quit Beyond Usual Care (95% CI)
Study 1
BNI+NRt+QL+booster	$860 ($825 - $953)	7.3% (3.3%-11.5%)	$11,814 ($7,521 - $25,349)
Study 2
Cost/BNI quit	$372 ($364 - $379)	4.5% (0.8%, 8.3%)	$8,184 ($4,255, $36,388)
Cost/NRT quit	$593 ($583 - $603)	6.4% (2.7%, 10.2%)	$9,209 ($5,782, $22,250)
Cost/Quitline quit	$284 ($268 - $381)	2.3% (−1.5%, 6.1%)	$12,496 ($4,522, infinity)
Cost/Textline quit	$337 ($335 - $340)	0.8% (−3.0%, 4.6%)	$44,484 ($7,292, infinity)
Add NRT to BNI	$374 ($364 - $384)	6.4% (2.7%, 10.2%)	$5,808 ($3,656, $14,146)
Add QL to BNI + NRT	$63 ($46 - $160)	2.3% (−1.5%, 6.1%)	$2,760 ($820, infinity)
BNI & NRT combined	$746 ($729 - $763)	11.0% (6.0%, 16.0%)	$6,791 ($4,564, $13,259)
BNI, NRT & QL combined	$808 (775 - $813)	13.3% (7.6%, 18.9%)	$6,100 ($4,043, $12,274)

Table 2.

Minutes and cost per participant and per quit based on as-available staff.

Treatment	Minutes/Participant	Cost/Participant (95% CI)	Cost/Quit Beyond Usual Care (95% CI)
Study 1
BNI+NRt+QL+booster	44.2	$227 ($192, $321)	$3,121 ($1,915, $6,965)
Study 2
Cost/BNI quit	13.7	$14 ($6, $21)	$307 ($102, $1,410)
Cost/NRT quit	18.2	$189 ($179, $199)	$2,929 ($1,839, $7,135)
Cost/Quitline quit	4.6	$19 ($2, $116)	$821 ($60, infinity)
Cost/Textline quit	4.6	$72 ($69, $74)	$615 ($1,533, infinity)
BNI & NRT combined	31.9	$203 ($190, $215)	$1,844 ($1,233, $3,602)
BNI, NRT & QL combined	36.5	$221 ($195, $322)	$1,669 ($1,052, $3,531)

Only BNI and NRT resulted in statistically significant increases in the quit rate in Study 2. Moreover, the effectiveness of BNI and NRT combined in Study 2 was 1.5 times that observed in Study 1. Consequently, the BNI-NRT combination had a much lower cost per quit relative to usual care ($6,791, 95% CI: $4,560 - $13,259) than the Study 1 intervention. In the bootstrap simulation, quitline referral had only an 88% probability of increasing quits, while the text booster had only a 65% probability. Hence the upper tail of the 95% CI for the cost per quit with these interventions is infinite. Despite the 12% chance that quitline referral is not effective, combining that intervention with BNI and NRT yields the lowest cost per quit and the most favorable confidence interval ($6,100, 95% CI: $4,043 - $12,274).

Because recruitment cost did not rise as intervention components were added, with dedicated staff, the lowest mean costs per quit were the incremental costs to add NRT and the quitline to BNI. With repurposed staff, although far fewer people might be served, the cost per quit would be much lower (Table 2). The pattern of cost per quit is similar with repurposed staff, except that the incremental cost from adding components is the same as the costs of the components in isolation.

Discussion

This paper produced cost-effectiveness estimates for two successful smoking interventions implemented in an ED.^9,10 Compared to the general population, ED patients have disproportionately low socioeconomic status, are more likely to smoke, and have less access to primary care. Thus, the study focused on a particularly vulnerable, understudied smoking population.

We used two distinct approaches for estimating intervention costs per participant and costs per quit. The first assumed dedicated staff and included costs of supervision, recruitment, idle time for cessation staff, and contracted services. The combined BNI, NRT, and quitline arms were the most cost-effective intervention with 13.3% effectiveness and a $6,100 cost per quit. This compares favorably with Piper et al.,¹³ who describe an intervention implemented in a primary care setting (NRT plus 3 hours of counseling) that was 10% effective and cost $7,800 per quit. This same combination of interventions plus booster calls, however, only increased quits by 7.3% in the first trial and cost $11,800 per quit.

Our second costing approach paralleled much of the literature and reflects a repurposed staffing philosophy. It ignored costs for supervision, costs of unsuccessful recruitment, and idle time between participants. It merely costed time spent delivering the intervention, plus medication and contract/donated service costs. In this context, BNI + NRT + quitline cost $221 per participant and $1,669 per quit. By comparison, Drouin et al.¹⁴ reported a $42 cost per participant and a $762 cost per quit (in 2018 dollars) among parents served in a pediatric care setting. That article reports the published costs per quit for five other smoking cessation interventions ranged from $554 to $4,674 (also in 2018 dollars), with two being less cost-effective than the ED-based approach analyzed in this study.

In addition to omitting supervisory costs, interventions by repurposed staff may be less well suited for EDs than interventions by dedicated staff. During busy hours when patients backlog, repurposed staff may lack capacity to recruit for and deliver a non-essential intervention. Reducing intervenor unavailability requires maintaining a large cadre of trained staff. Because they intervene more regularly and without other pressing duties, dedicated staff also may deliver the intervention with greater fidelity.

Limitations

The major limitation of this study reflects the limitations of the underlying clinical trials. Foremost among those, the quitline referral factor may have been underpowered given the size of its effect. It induced 2.3% of participants to quit, but the standard deviation of that reduction equaled 85% of the mean. Omitting the quitline, the cost effectiveness ratio is slightly less favorable: $6,791 per quit with dedicated staff and $1,844 per quit with as-available staff. Challenges in allocating recruitment and standby time also reduce the robustness of this study’s cost estimates.

Further, EDs are not necessarily set up to deliver behavioral interventions. This paper omits costs for an ED to establish the necessary delivery pathway. The interventions evaluated excluded some cost-effective measures, notably, peer recovery coaches.¹⁵ Implementing additional support strategies might affect the cost-effectiveness of the interventions.

Conclusions

EDs can be an important location for interventions designed to reduce smoking. The studies described demonstrate that interventions based on dovetailing treatments including BNI, NRT, and quitline referral can be effective and cost-effective in reducing smoking among ED populations. The assumptions made when estimating cost-effectiveness vary in the literature, and care is needed when comparing reported intervention costs across studies.