Introduction 1 2 7 4 5 8 11 12 13 14 14 Methods The model 15 16 17 17 1 1 1 1 12 14 17 1 1 17 1 Fig. 1 The circles  Scenarios 17 17 1 2 after 1 18 1 18 1 2 1 2 Table 1 Mean (SE) probabilities to move between disease states in the base-case scenario. (These probabilities were also used in scenario 3. In scenario 2 the subsequent cycle probabilities for year 1 were applied to years 2 to 5. The cycle length was 8 days for the very first cycle and 1 month for of all subsequent cycles)   Tiotropium Salmeterol Ipratropium Moderate Severe Very severe Death Moderate Severe Very severe Death Moderate Severe Very severe Death First cycle probabilities  Moderate 0.906 (0.018) 0.092 (0.018) 0.001 (0.002) 0.001 (0.001) 0.898 (0.033) 0.100 (0.033) 0.001 (0.003) 0.001 (0.001) 0.738 (0.092) 0.257 (0.091) 0.004 (0.013) 0.001 (0.002)  Severe 0.259 (0.017) 0.715 (0.018) 0.025 (0.006) 0.001 (0.001) 0.201 (0.028) 0.765 (0.030) 0.033 (0.013) 0.001 (0.002) 0.102 (0.028) 0.841 (0.034) 0.056 (0.022) 0.001 (0.002)  Very severe 0.010 (0.005) 0.340 (0.024) 0.648 (0.024) 0.002 (0.002) 0.001 (0.003) 0.301 (0.042) 0.696 (0.042) 0.002 (0.004) 0.005 (0.011) 0.220 (0.066) 0.773 (0.067) 0.002 (0.007) Subsequent cycle probabilities, year 1  Moderate 0.957 (0.010) 0.040 (0.010) 0.003 (0.003) 0.001 (0.001) 0.929 (0.025) 0.066 (0.024) 0.005 (0.007) 0.001 (0.002) 0.924 (0.050) 0.073 (0.049) 0.003 (0.010) 0.001 (0.004)  Severe 0.023 (0.006) 0.952 (0.009) 0.023 (0.006) 0.002 (0.002) 0.023 (0.011) 0.916 (0.020) 0.059 (0.017) 0.002 (0.003) 0.013 (0.011) 0.948 (0.021) 0.037 (0.018) 0.002 (0.004)  Very severe 0.001 (0.002) 0.045 (0.012) 0.947 (0.013) 0.008 (0.005) 0.006 (0.008) 0.036 (0.019) 0.951 (0.023) 0.008 (0.009) 0.003 (0.009) 0.025 (0.026) 0.964 (0.031) 0.008 (0.015) Subsequent cycle probabilities, years 2 to 5  Moderate 0.984 (0.007) 0.015 (0.007) 0.000 (0.000) 0.001 (0.001) 0.981 (0.017) 0.019 (0.017) 0.000 (0.000) 0.001 (0.003) 0.978 (0.033) 0.022 (0.033) 0.000 (0.000) 0.001 (0.005)  Severe 0.000 (0.000) 0.990 (0.004) 0.008 (0.003) 0.002 (0.002) 0.000 (0.000) 0.987 (0.008) 0.011 (0.007) 0.002 (0.003) 0.000 (0.000) 0.986 (0.012) 0.011 (0.011) 0.002 (0.005)  Very severe 0.000 (0.000) 0.000 (0.000) 0.992 (0.005) 0.008 (0.005) 0.000 (0.000) 0.000 (0.000) 0.992 (0.007) 0.008 (0.007) 0.000 (0.000) 0.000 (0.000) 0.992 (0.011) 0.008 (0.011) Table 2 Mean (SE) exacerbation probabilities in the base-case scenario. (These probabilities were also used in scenario 2. In scenario 3 the probabilities of ipratropium were also applied to tiotropium and salmeterol during years 2 to 5)   Probability of experiencing an exacerbation Probability that the exacerbation will be severe, given that there is an exacerbation Tiotropium Salmeterol Ipratropium Tiotropium Salmeterol Ipratropium Moderate 0.051 (0.004) 0.057 (0.013) 0.080 (0.020) 0.097 (0.024) 0.030 (0.031) 0.267 (0.114) Severe 0.075 (0.003) 0.089 (0.011) 0.097 (0.013) 0.136 (0.018) 0.138 (0.033) 0.188 (0.041) Very severe 0.096 (0.005) 0.104 (0.016) 0.102 (0.022) 0.192 (0.027) 0.207 (0.048) 0.186 (0.062) Input data: baseline distribution of patients among disease states 19 Input data: death state 20 1 20 1 Input data: utilities n 21 22 23 24 Input data: costs 3 25 26 Table 3 HU HU RP GP ICU exa  Unit costs Moderate COPD Severe COPD Very severe COPD HU per patient per year Total costs HU per patient per year Total costs HU per patient per year Total costs Visit RP in hospital 79.85 0.27 22 (1) 0.44 35 (2) 0.63 50 (3) Visit RP outside hospital 23.77 0.38 9 (1) 0.57 14 (1) 0.65 15 (1) Visit GP 8.35 2.59 22 (1) 2.88 24 (1) 3.27 27 (1) Thorax X-ray 19.67 1.03 20 (1) 1.18 23 (1) 1.49 29 (1) ECG 22.53 0.80 18 (1) 0.87 20 (1) 1.15 26 (1) Spirometry 40.63 0.55 22 (1) 0.66 27 (1) 0.92 37 (2) Blood analyses 19.67 1.41 28 (1) 1.43 28 (1) 1.58 31 (1) Blood gases 27.25 0.33 9 (1) 0.56 15 (1) 0.67 18 (1) Influenza vaccination 5.47 0.48 3 (1) 0.64 4 (1) 0.72 4 (1) Theophylline 0.18 122.06 22 (8) 161.77 30 (4) 159.07 29 (5) Mucolytics 0.30 39.74 11 (3) 48.31 14 (2) 80.60 24 (4) Oral corticosteroids 0.29 21.54 6 (3) 23.73 7 (1) 78.48 23 (6) Inhaled corticosteroids 0.61 224.84 138 (12) 224.84 138 (15) 292.00 179 (22) Oxygen 3.73 21.32 79 (14) 44.26 165 (30) 77.87 290 (52) Other   20 (3)  44 (4)  34 (5) Total costs per patient per year   430 (24)  587 (34)  818 (58) Exacerbations Unit costs Non-severe exacerbation Severe exacerbation Study medication costs per day HU per exa Total costs HU per exa Total costs a 1,284 – – 0.29 374 (291) Tiotropium 1.80 a 368 – – 4.16 1,529 (307) Salmeterol 1.20 Emergency room 115 – – 0.94 108 (11) Ipratropium 0.19 Visit GP 8.35 1.64 14 (1) 1.00 8 (3)   Visit RP in hospital 79.85 – – 0.52 42 (10)   b 5.00 11.02 55 (7) 7.52 38 (7)   b 0.29 2.69 1 (1) 4.98 1 (1)   b 0.61 7.01 4 (1) 3.71 2 (1)   b 3.73 2.05 8 (3) 5.32 20 (2)   Other   1 (1)  1 (1)   c s74.09    13 (2)   Total costs per exacerbation   83 (7)  2,136 (425)   For medications and oxygen HU is expressed in number of days during which the medication or oxygen was used a b c 25 27 27 The NHS covers the costs of absence from work due to illness from the 16th day of sick leave onwards. The NHS pays 60% of gross salary during days 16 to 20 and 75% from day 21 onwards. When we were calculating costs from the NHS perspective, these costs were added to the costs of all severe exacerbations with sick leave longer than 15 days (approximately 7% of all severe exacerbations). The costs from the societal perspective included the costs of lost production during all days of absence from paid work. The resulting mean (SE) annual costs of maintenance therapy from an NHS perspective were €430 (€24) for a patient with moderate COPD, €587 (€34) for a patient with severe COPD and €818 (€58) for a patient with very severe COPD. The corresponding values from the societal perspective were €429 (€21), €586 (€34) and €816 (€58). The mean (SE) costs of exacerbations from the NHS perspective were €83 (€7) for a non-severe exacerbation and €2,176 (€425) for a severe exacerbation. From a societal perspective these costs were €121 (€7) and €3,912 (€543), respectively. COPD severity To investigate the impact of COPD severity on the cost effectiveness of the bronchodilators we ran sensitivity analyses with the base-case scenario, assuming that, at the start of the model, 100% of the patients had either moderate, severe or very severe COPD. Discounting 28 Probabilistic sensitivity analysis 17 29 30 30 31 32 curve frontier C E C E 33 Results Health outcomes 4 Table 4 QALY Tio Salm Ipra exa CE-ratio exa-free  Costs 4 4 Cost effectiveness 4 2 32 2 2 2 Fig. 2 curves in grey in black  Societal perspective From the societal perspective, overall mean (SE) costs increased in all treatment groups to €6,574 (€321) for tiotropium, €6,125 (€541) for salmeterol and €5,545 (€720) for ipratropium. However, the savings in exacerbation costs generated by tiotropium and salmeterol compared with ipratropium were higher from a societal perspective than from the NHS perspective. Consequently, from the societal perspective, the difference in overall costs between the treatment groups was reduced. Also, the incremental costs per exacerbation-free month were reduced to €308, for the comparison tiotropium versus salmeterol, and €1,375 for the comparison salmeterol versus ipratropium. The incremental cost per QALY were reduced to €3,483, for the comparison tiotropium versus salmeterol, and €35,158 for the comparison salmeterol versus ipratropium. Tiotropium had the highest expected net benefit for any value of the cost-effectiveness threshold above €547 per exacerbation-free month and above €7,076 per QALY. Below these values ipratropium is preferred. Impact of COPD severity When the model was run separately for patients with either moderate, severe or very severe COPD, it showed that the threshold value above which tiotropium had the highest expected net benefit increased with the severity of COPD. The threshold values for the costs per QALY above which tiotropium became the preferred option were €7,600 for moderate COPD, €8,800 for severe COPD and €12,500 for very severe COPD. Below these values ipratropium was preferred. Tiotropium had the highest expected net benefit when the ceiling ratios for cost per exacerbation free month were €560 for moderate COPD, €700 for severe COPD and €1,200 for very severe COPD. Below these values ipratropium had the highest expected net benefit. Discount rates Because discounting affects both health outcomes and costs, the effect of discounting in this study was small. Analyses based on discount rates of 3% and 0% showed almost similar cost-effectiveness ratios. Discounting health outcomes at a lower rate than costs led to cost-effectiveness ratios that were slightly more in favour of tiotropium. Discussion Bronchodilators form the main therapy for the symptomatic relief of respiratory symptoms in COPD patients. In this study we have constructed a decision analytical model to synthesise clinical trial data on the effectiveness of bronchodilator treatment. We have shown how scenario analyses can be used to extend the time horizon of the cost-effectiveness study beyond that of the clinical trials. In addition, we have shown how a model can be populated with country-specific data to obtain estimates of the cost-effectiveness of bronchodilators in the Spanish setting. 17 19 21 25 26 34 The base-case scenario showed that tiotropium was associated with an approximately 16% reduction in exacerbations when compared with salmeterol. Salmeterol was associated with 12% reduction in exacerbations, when compared with ipratropium. Differences between the three treatment groups in terms of QALYs were small. That was expected, given the 5-year time horizon and treatments that do not directly affect survival. The distribution of dots on the CE-planes showed that the higher effectiveness of salmeterol over ipratropium was associated with more uncertainty than the higher effectiveness of tiotropium over salmeterol, because, for the latter comparison, a larger proportion of the dots was found on the right side of the CE plane. 35 36 17 37 38 39 40 41 When the cost-effectiveness analysis was conducted from a societal perspective, the threshold value above which tiotropium is the preferred option is reduced considerably to €547 per exacerbation-free month and €7,076 per QALY. This improved cost effectiveness is primarily related to increasing exacerbation costs, because of the inclusion of the costs of production losses. Hence, a reduced exacerbation rate is associated with higher savings. The NHS perspective also included sick leave benefits paid for by the NHS from day 16 onwards. However, this applies only to severe exacerbations where it was less than 1% of its costs. In contrast, from a societal perspective, costs of production losses are 10% of the costs of a severe exacerbation and 30% of the costs of a non-severe exacerbation. 42 43 In conclusion, our model has demonstrated that tiotropium is the treatment with the highest expected net benefit, if decision makers can afford to spend additional budget to gain additional health benefits. The threshold value of the costs per QALY at which tiotropium becomes the preferred treatment is well within acceptable limits (i.e. €8,157 from the NHS perspective and €7,076 from the societal perspective).