This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.

The study evaluated the use of a tobacco cessation intervention for mothers of small children.

Primary prevention.

Cost-utility analysis.

The study population comprised mothers with children aged 0 to 6 years who were daily smokers. No other inclusion criteria or exclusion criteria were reported in this paper.

The setting appears to have been primary care. The economic study was carried out in Sweden.

The effectiveness data were gathered between 1987 and 2003. The resource use data were gathered between 1994 and 2001. The price year was 2000.

The effectiveness data were derived from a review of published studies.

A Markov model was developed to estimate the incremental costs of the smoking cessation intervention, compared with the do-nothing option, using Data 3.5 software (Treeage Inc.). The health states of the model were lung cancer, chronic pulmonary disease (COPD), cardiovascular disease including coronary heart disease and stroke, and death. A probabilistic analysis was carried out using Monte Carlo simulation. Each cycle was 1 year long. The simulation terminated either at death or at age 85 years, except when costs in added years were considered in the analysis (a lifetime horizon was then used). Details of methods and parameters distributions were reported elsewhere (Johansson, see 'Other Publications of Related Interest' below for bibliographic details).

The main outcomes estimated from the review of the literature were the risks of disease and death associated with lung cancer, COPD and acute myocardial infarction (AMI) for both smokers and quitters. Age- and gender-specific disease risks, conditional on smoking status and years since quitting, were used for the analysis.

Not stated.

Not stated.

Not stated.

Not stated.

Approximately 8 studies were included in the review.

Not stated.

No differences between the primary studies were investigated.

The risk of developing a lung cancer was 0.12 (standard deviation, SD=0.33) for smokers and 0.01 (SD=0.10) for quitters.

The risk of developing COPD was 0.08 (SD=0.28) for smokers and 0.03 (SD=0.17) for quitters.

The risk of suffering an AMI was 0.05 (SD=0.22) for smokers and quitters.

For smokers, the risk of death associated with these conditions was 0.10 (SD=0.30) for lung cancer, and 0.03 (SD=0.17) for COPD and AMI.

For quitters, the risk of death associated with these conditions was 0.01 (SD=0.09) for lung cancer, 0.01 (SD=0.11) for COPD and 0.03 (SD=0.16) for AMI.

These data formed the principal effectiveness parameters used in the analysis. The authors only reported selected estimates for women aged 35 to 39, until the age of 85 years.

The outcome measures were life-years saved (LYS) and quality-adjusted life-years (QALYs). A variety of sources were used to obtain disease-related death risks for both quitters and smokers. For other causes of death, estimates from the Swedish National Death Registry estimates were used. The number of LYS was calculated until age 85 and predicted by the model. Utility values were taken from a Swedish representative population study, (Burstrom et al. 2001, see 'Other Publications of Related Interest' below for bibliographic details). The method of valuation was not reported in this paper. The outcomes were discounted at a rate of 3%.

The direct costs reported were those of the health service. The key resource use categories included were initial disease-specific medical treatment costs (i.e. those accumulated during 6 months after the assumed first admission to hospital), annual disease-specific medical treatment costs, costs of death (i.e. those accumulated during 6 months before death), and costs related to the intervention. The latter (intervention-related costs) included personnel, overheads, campaign material and publicity, participant support measures and the prize ceremony. Resource use was not reported separately from the costs. Discounting was carried out at a rate of 3%. The costing appears to have been based completely on actual data. The price year was 2000.

The costs were treated stochastically. The methods used for the statistical analysis of the costs were not reported in this paper.

Indirect costs were also considered. The cost analysis included income loss because of absence from work and morbidity productivity costs before the age of 65 years. Participant time spent attending the smoking cessation meetings was also included. Resource use was not reported separately from the costs. Discounting was carried out at a rate of 3%. The costing appears to have been based completely on actual data. The price year was 2000.

Swedish kroner (SEK). The conversion rate was Euro 1 = SEK 8.45.

Several univariate sensitivity analyses were performed. These varied all model parameters over a range of values (+100% to -50%, approximately). Two sets of multivariate analyses were also performed. These varied the risk of disease and death (high risk: death and disease risk +100%, risk fraction +0.1; low risk: death and disease risk -50%, risk fraction -0.1). They also investigated the addition of an increment of +25% on all costs for the high-risk group and a reduction of -25% on all costs for the low-risk group. The method used to select the ranges was not specifically stated, but assumptions may have been the basis of the selection.

The model estimated that the difference between smokers and quitters was between 0.34 (age group 15 to 19 years) and 0.55 (age group 30 to 34 years) QALYs gained until the age of 85 years. The undiscounted LYS for smokers and quitters differed by 1 to 2 years. The authors did not specify the incremental benefits of the intervention compared with the control strategy.

Based on the results of the "Quit Smoking Gals" programme undertaken in one of the health care districts in Stockholm County Council, 238 women were recruited, of whom 34 remained tobacco-free for 12 months. The number of LYS was estimated to be 20, while the number of QALYs gained because of the intervention was 16.

The total costs for a female smoker were estimated to be between SEK 100,000 and SEK 180,000, depending on age group. For a quitter, these costs were between SEK 80,000 and SEK 150,000.

The cost analysis revealed the savings associated with smoking cessation amounted to about SEK 30,000 per female quitter.

If disease-related morbidity productivity costs were excluded, the savings per quitter were approximately SEK 20,000.

The authors did not specify the incremental costs of the intervention compared with the control strategy.

Based on the results of the "Quit Smoking Gals" programme undertaken in one of the health care districts in Stockholm County Council, the intervention costs amounted to SEK 267,000, the cost per participant was SEK 1,100 and the cost per quitter was SEK 7,850. The cost per undiscounted LYS was SEK 4,100.

The "Quit Smoking Gals" intervention led to societal cost-savings of SEK 830,000. These savings ranged from SEK 2,620,000 to SEK 420,000 according to the discount rate.

The sensitivity analysis showed that the parameter with the highest impact on the results was the discount rate.

The costs and benefits were not combined.

The "Quit and Win" contest was associated with cost-savings and health gains.

A do-nothing option was used as the comparator. This seems to have been appropriate for this type of analysis. You should decide whether this could be a valid comparator in your own setting.

The effectiveness data were derived from a review of published studies, but a systematic review of the literature was not undertaken. Although this is common practice with models, it does not always ensure that the best data available are used in the model. Details of the primary studies and the methods used to combine the primary estimates were not provided. The effectiveness evaluation was designed as a before-and-after study, which, as the authors acknowledged, entails less scientific rigour than other evaluation designs. In addition, the authors did not justify their choice of the type of patient population (mothers with children aged 0 to 6 years who were daily smokers). Uncertainty around the outcome parameters was investigated in sensitivity analyses.

The authors used the numbers of LYS and QALYs gained from smoking cessation intervention as their summary measures of health benefit. These measures represent natural outcomes from a preventive programme and enable comparisons with similar vaccination evaluations. The outcomes results were left disaggregated and no incremental benefits of the smoking cessation programme over the do-nothing approach were explicitly presented.

The analysis of the costs was performed from a societal perspective. It appears that all the relevant categories of costs have been included. Although the costs were reported separately to other model parameters, they were presented in an aggregated manner. Details of the unit costs and the quantities of resources used were not given, which limits the transferability of the study results to other settings. The costs were treated deterministically and were not subjected to any statistical tests. All cost components were explored in the sensitivity analysis. No justification was given for the ranges used, and it was not possible to determine whether they were appropriate. The source of the cost data was reported for each item. The price year and conversion rate were reported, which will aid reflation exercises in other settings. Discounting was conducted and was relevant because of the long time horizon. No incremental costs of the smoking cessation programme compared with the do-nothing option were explicitly presented.

Although the authors highlighted the relevance of conducting a cost-utility analysis using generic outcome measures, no incremental analysis (incremental cost-effectiveness ratios) was presented. The way in which the authors reported the mean costs and outcomes for the smoker and quitter status was confusing: one could assume these were equivalent to the intervention and control strategy, but it was not clear how the probability of quitting smoking after taking part in the "Quit-and-Win" contest was implemented in the model. The authors discussed the results of the analysis in the context of the published literature. The issue of generalisability to other settings was not addressed. Limitations reported by the authors included the potential underestimation of costs for diseases other than stroke, the overestimation of medical treatment costs, and the fact that true opportunity costs were not reflected because the administrative system was used to assess medical treatment costs.

The authors stated that the model results show considerable health gains and cost-savings resulting from tobacco cessation among women. The authors highlighted that the construction of an optimal mix of tobacco control policies would demand economic evaluations of a wide range of tobacco control programmes.

None stated.

Johansson P. A model for economic evaluation of smoking cessation interventions. Available from: URL: www.folkhalsoguiden.se/templates/Publication_1045.aspx

Burstrom K, Johannesson M, Diderichsen F. Health-related quality of life by disease and socio-economic group in the general population in Sweden. Health Policy 2001;55:51-69.