Overall description

We performed a cost-utility analysis of FAHSAI Clinic compared with usual care in a population of Thai smokers with CVD disease, from a societal perspective. The effectiveness and cost of FAHSAI Clinic were primarily based on data from a multicenter prospective observational study of 24 FAHSAI clinics across 21 provinces of Thailand⁸. The cohort study included 2041 participants who were first commencing the smoking cessation program in the multidisciplinary clinics and had a mean age of 44.56 years⁹. We used an annual cycle length. As per Thai HTA guidelines⁹, we discounted QALYs and costs at 3% per year¹⁰. This work is reported in accordance with the Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022)¹¹ (Supplementary file) and complies with the Guidelines for health technology assessment (HTA) in Thailand (Second edition)⁹.

Model structure

We developed a Markov model to simulate lifetime costs and quality-adjusted life years (QALYs) of a hypothetical cohort of smokers receiving smoking cessation services from either FAHSAI Clinic or usual care over a lifetime horizon, i.e. 50 years, as this accounts for all costs and medically important events (including the risk of a cardiovascular event and death). Our target population included Thai men (93%) and women (7%) aged 35 years, who regularly smoke 10–20 cigarettes per day. An age of 35 years was selected as it approximates the average age of Thai smokers¹².

After commencing the smoking cessation program or usual care, a patient in the stable CHD/stroke state could remain in this state or transition to an acute disease exacerbation state. Patients in acute disease exacerbation states could either remain in this state or transition to a state of stable secondary stroke/CHD or stable CVD. The transition to the next CVD health state is assumed to be irreversible. Patients could also transition to the states of lung cancer, COPD, mouth cancer or death at any point in the model. As patients moved through different health states cycles, they accrued direct healthcare costs, life-years, and QALYs.

This model has two facets representing the two cohorts of the starting population: patients with stable coronary heart disease (CHD) (Cohort A) and patients with stable stroke (Cohort B), each consisting of 9 mutually exclusive, discrete health states. The health states within the CHD cohort facet of the model include: 1) stable CHD, 2) acute secondary CHD event, 3) stable secondary CHD, 4) acute stroke in CHD population, 5) stable CVD, 6) mouth cancer, 7) lung cancer, 8) chronic obstructive pulmonary disease (COPD), and 9) death (Figure 1). The health states within the stroke cohort facet of the model are: 1) stable stroke, 2) acute secondary stroke event, 3) stable secondary stroke, 4) acute CHD in stroke population, 5) stable CVD, 6) mouth cancer, 7) lung cancer, 8) COPD; and 9) death (Figure 2). The age-stratified transition probabilities used in the CHD and stroke cohort models are presented in the Supplementary file Tables 1 and 2, respectively. In this model, there were 3 main assumptions: 1) each smoker only receives one smoking cessation therapy during their lifetime; 2) the risk of smoking related diseases was dependent on the current health state or a limited set of previous states; and 3) smokers will suffer from at most one smoking-related disease during their lifetime.

Figure 1

Markov model of the coronary heart disease cohort health states applied for cost-effectiveness analysis

Figure 2

Markov model of the stroke cohort health states applied for cost-effectiveness analysis

Input parameters

Transition probabilities

Risks of developing smoking-related diseases were based on published studies. The risk of CHD among current and former smokers were derived from the Kaiser Permanente Medical Care Program cohort study, which had a mean follow-up length of 6.1 years¹³. The risk of stroke and COPD were based on CHD statistics published by the British Heart Foundation Statistics database¹⁴. The risk of lung cancer was obtained from a nationwide American Cancer Society prospective cohort¹⁵, while the risk of oral cancer was based on results of a Japanese prospective cohort¹⁶.

The model accounts for the reduced risk of developing smoking-related conditions among smokers who quit successfully (ex-smokers) which were compared with continuing smokers (smokers). The risks smoking-related disease and disease specific mortality were derived from targeted literature searches. Sex-specific relative risks of death from CHD, stroke, and COPD in current and former smokers were derived from the Kaiser Permanente Medical Care Program cohort study¹³. The risk of death from oropharyngeal and lung cancer by smoking status were obtained from Japanese cohort studies^16,17. The relative risks of incident smoking-related disease were calculated by multiplying the number of Thai smokers by sex-specific absolute risks of smoking-related disease and the relative risk of disease-specific mortality and then divided by the total number of smokers. Transition probabilities from a CHD or stroke health state to death were derived by multiplying Thai age-specific mortality rates from the National Statistical Office of Thailand by the relative risks of disease specific mortality¹⁸.

The effectiveness of FAHSAI Clinic was based on 6-month continuous smoking abstinence rate (CAR) reported in the concurrent multicenter prospective cohort study of FAHSAI Clinics. In this study, 13.8% of participants enrolled in the FAHSAI Clinic abstained from smoking at 6 months of follow-up⁸. As FAHSAI Clinics are well established across Thailand and considered as standard care to all Thai smokers, it would not be feasible and ethical to include a control group who did not receive smoking cessation service as a comparison. We assumed that 5% of patients receiving usual care for smoking cessation abstain from smoking at 6 months of follow-up based on opinion of clinical experts (SW, SR, AT). As this quit rate follows a binary distribution, its standard error can be estimated from:

where σ is a standard deviation, n is the number of smokers that participated in the study, and p is the quit rate and q=1-p.

Cost and health utility values

Costs were calculated from a societal perspective. In addition to smoking cessation treatment costs, other costs considered in the model encompassed other medical costs (including medication and use of health services). As per the HTA guidelines for Thailand, indirect costs were not included as we have accounted for health utilities in our model⁹. The costs for CHD and stroke in the first and subsequent years came from cost studies and expert opinion^19-22. Direct medical and non-medical costs for COPD, oral and lung cancer were also estimated from Thai studies^21,23,24 and data collected at the Maharajnakorn Chiang Mai Hospital and validated based on expert opinion of research team clinicians (SW, SR, AT). Cost data were estimated in Thai Baht and converted to US$ using a rate of THB 31.25 to US$1. All costs were adjusted to 2020 dollars using the Consumer Price Index²⁵. Baseline utility values were derived from a cost-effectiveness analysis of Belgium smokers, an interim determination of health gain from oral cancer and precancer screening done in the UK, and from a prospective quality-of-life survey on advanced non-small-cell lung cancer patients in Europe, Canada, Australia, and Turkey^26-28. The model input parameters are summarized in Supplementary file Table 3.

Sensitivity analyses

We conducted a series of sensitivity analyses to assess the robustness of our results. One-way sensitivity analysis was performed for baseline event rates, relative risk estimates, costs, utility values, and the discount rate used in the model. We also conducted a probabilistic sensitivity analysis (PSA) for all parameters in the model using a Monte Carlo simulation technique with 1000 iterations, which was used to create a cost-effectiveness acceptability curve. Beta distributions were used for transition probabilities and utility values, the log-normal distribution was used for relative risk, and the gamma distribution was assigned to cost data. The PSA results were used to create cost-effectiveness acceptability curves, which show the probability of FAHSAI Clinic being cost-effective over a range of willingness-to-pay thresholds. As there were large difference in the 6-month CARs obtained in the per-protocol and intention to treat (ITT) analyses⁸, a scenario analysis was conducted to explore variation in our results that resulted from the use of effectiveness estimates measured per-protocol versus ITT.

Limitations

This study has some limitations. First, because the effectiveness data were obtained from a cohort study, it may be subject to both selection bias and confounding commonly present in observational studies; and the COVID-19 pandemic may have further compromised study findings. For example, the cohort study used two measurements for smoking outcomes: a self-reported questionnaire and an exhaled carbon monoxide (CO) test as a biochemical validation method. The exhaled-CO test, however, was prohibited to prevent viral transmission. Additionally, the pandemic lead to reduced follow-up, resulting in decreased responses from study participants; among participants with CVD only 33% had complete outcome data. Our base case utilized an ITT analysis to obtain conservative estimates of the program efficacy and despite the use of these estimates in our model, the FAHSAI Clinic was found to be cost-effective. Further, there was a large difference in the 6-month CAR in the per-protocol and ITT analyses, 41.77% and 13.81%, respectively. Depending on available resources, each multidisciplinary FAHSAI Clinic may offer differing services and treatments. For example, only 68.40% of participants in their cohort were receiving any pharmacotherapies. As participants who had not used any pharmacotherapies had lower likelihoods of successfully quitting smoking, compared to those who had, it is evident that these variations in the services and treatments offered may have attenuated the program’s overall effectiveness. As such, we ran a scenario analysis using the per-protocol efficacy estimates and the FAHSAI clinic remained dominant over usual care over a 50-year time horizon, with less costs associated with gains in QALYs compared to the base-case analysis.

The second limitation was that because smoking cessation clinics are considered the standard care for Thai smokers, the current study compared the cost-effectiveness of FAHSAI Clinic to usual care but it was not possible to obtain the effectiveness of the clinic compared to self-quit attempt in analyses. There was also a lack of comparative evidence on subsequent quit attempts in participants receiving care at FAHSAI Clinics and under usual care. The model assumed that smokers do not attempt smoking cessation again after a first failed attempt until death, although several quit attempts may be required before they are successful. In addition, despite the possibility for smokers to develop multiple comorbidities, our model assumed that smokers would suffer from at most one smoking-related disease in their lifetime. However, this would likely underestimate the harm induced by smoking and result in an attenuation of the observed benefit from the smoking cessation interventions. Further, as we performed English-language searches, Asian-based data were limited and therefore the relative risks of death from smoking related disease and utility values used in our model were derived from studies done in the US, Europe, and Japan, which may reduce generalizability to Thai smokers.

Considerations and implications

An important consideration was that this study considered multidisciplinary clinical smoking cessation as an intervention for tobacco control. To date, much of the research on smoking cessation interventions compare individual strategies or a combination of two strategy types. A systematic review and meta-analysis of randomized controlled trials, found that multiple behavioral change strategies were more effective than usual care and brief advice interventions alone³², however, the effectiveness rates reported were lower than those observed from the FAHSAI Clinic. In reality, population-wide policy, systems, and environmental changes have also been shown to be highly efficient and effective at reaching many people resulting in increased motivation to quit and demand for tobacco dependence treatment³³. However, tailored programs such as the FAHSAI Clinic tend to be more effective compared with these standardized interventions, thereby providing greater effects on health behaviors³³.

Our findings have policy implications to support smoking cessation services in Thailand. Despite being implemented in Thailand in 2010, the FAHSAI Clinic has not yet been included in the National Health Security Office (NHSO) health benefit package. Our results demonstrated its cost-effectiveness compared to usual care among Thai smokers with CVD, thereby providing strong support for policy makers to consider including FAHSAI as part of their benefit package.