INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: cost effectiveness analysis; early treatment; asthma; COPD; antiinflammatory therapy

The benefits of treating asthma with inhaled corticosteroids are well established (1, 2). Inhaled corticosteroids are the cornerstone of asthma treatment and occupy a prominent position in international guidelines (3, 4). The role of inhaled corticosteroids in the treatment of chronic obstructive pulmonary disease (COPD) is less clear (5). A number of studies demonstrated significant advantages, but the benefits in patients with COPD were less perceptible than those in patients with asthma (6, 7). Recently, three large randomized clinical trials have demonstrated that in the long term, inhaled corticosteroids have only limited effect on the forced expiratory volume in 1 s (FEV₁) (8). Two of the studies measured significant improvement in FEV₁, mainly during the first 6 mo of treatment, but the annual decline in FEV₁ was disappointing in the long term (8, 9). The third study on patients with COPD recruited from the general population did not find any clinical benefit (10). However, the significant improvements in (the severity of) exacerbations (9, 11) and health state (9, 12) indicated that treatment for COPD with inhaled corticosteroids did result in perceptible benefits, presumably via some as yet undefined antiinflammatory effect (13).

Over the past decade, treatment with inhaled corticosteroids has gradually expanded to include milder cases of asthma (2). Studies showed significant results with respect to airway inflammation, lung function, and reduction in hospital admissions in patients with newly diagnosed mild to moderate asthma (14). The rationale behind this shift was not only to treat patients with milder asthma, but also to start treatment at the earliest possible stage, in order to prevent irreversible loss of lung function and thus the transition from intermittent asthma to chronic persistent asthma. Two studies indicated that early treatment for asthma might prevent irreversible loss of lung function and thus might prevent airway remodeling (17, 18). A rationale for early intervention in COPD was provided in the EUROSCOP study (8). This study revealed that the effect of treatment with inhaled corticosteroids was associated with smoking history (8): the subgroup of patients with a smoking history of less than the median number of pack-years who had received inhaled corticosteroids showed a slight but significantly smaller annual decline in lung function than their counterparts in the placebo group. Hence, in this subgroup of patients with COPD inhaled corticosteroids may also prevent irreversible loss of lung function.

The literature substantiates the common belief that from a theoretical perspective, early medical intervention may be worthwhile in COPD and asthma. Early intervention with inhaled corticosteroids might prevent irreversible loss of lung function, preserve or improve quality of life, and may lead to substantial financial savings in the long term. In this study, which forms part of the Detection, Intervention and Monitoring of COPD and Asthma program (DIMCA), the cost effectiveness of early intervention with fluticasone propionate 250 µg twice daily was assessed in subjects with objective signs of obstructive airway disease, who were undiagnosed prior to the study. In this comprehensive analysis, we addressed the clinical effects, effects on quality of life, subject's preferences, and economic consequences.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject Selection

Subjects participating in the screening and first 12 mo of the monitoring phase of the DIMCA program (19) were selected for early fluticasone intervention if they met one of these criteria:

Obstructionat least two of the first four quarterly FEV₁ measurements less than the predicted value minus 2 SD;

Hyperresponsivenessa concentration of histamine provoking a 20% fall in FEV₁ (PC₂₀) < 2 mg histamine/ml and FEV₁ reversibility > 15% of the predicted value;

Decline in lung functionan annual decline in FEV₁ > 80 ml/yr and a PC₂₀  8 mg histamine/ml and/or FEV₁ reversibility  10% of the predicted value.

Trial Design

Subjects were randomized to fluticasone propionate 250 µg or placebo twice a day per pMDI. Salbutamol 400 µg per Rotadisk was rescue medication. Subjects were seen at the lung function laboratory at baseline, 6 and 12 mo, and at their general practice at 3 and 9 mo. This double-blind trial and the preceding detection program were approved by the Ethics Committee of the University Lung Centre, University of Nijmegen, The Netherlands.

Clinical Outcomes

Pre- and postbronchodilator FEV₁ was measured every 3 mo, according to ATS guidelines (20). A Microspiro HI-298 was used at the laboratory (21). A hand-held spirometer (Microplus, SensorMedics) was used at the general practice. Predicted values were calculated using the method of Quanjer and coworkers (22). Reversibility was defined as change in FEV₁, 15 min after inhalation of 800 µg salbutamol by Volumatic, relative to the predicted value (in %). At baseline and every 6 mo, a histamine provocation test was performed according to Cockcroft and coworkers (23). Subjects were allergic if they tested positive for at least one of 17 allergens at baseline. A symptom score (range 0-4) was calculated on the basis of the number of symptoms (cough, shortness of breath, wheeze, and productive cough), recorded weekly on a diary card. During a symptom-free week the symptom score was 0. The subjects indicated on the diary card if their symptoms had worsened. During an episode-free week the symptoms had not worsened. Exacerbations were episodes that required medical attention. During an exacerbation, at least two of the following three criteria had to be present: (1) episode with increased (productive) coughing and/or dyspnea and/or wheezing, (2) change in sputum color, or (3) increased use of bronchodilatory drugs (24).

Resources and Cost

Health care resource use was recorded by the general practitioner (GP) and verified by an audit of the subjects' medical records and a standardized 3-mo interview with each subject. On a diary card, subjects recorded nonprescription medication and the number of days they were unable to do paid or unpaid work (indirect cost). Resource use was multiplied with 1998 unit prices (25). Medication cost included pharmacy cost. Indirect cost was valued using the Human Capital Approach. Dutch guilders were converted into US dollars using 1998 purchasing power parities (1 US$ = NLG 2.08).

Cost Effectiveness Ratios

We calculated (1) the cost per quality-adjusted life year (QALY), (2) the cost per subject with a minimal clinically important difference (MCID) in the Chronic Respiratory Questionnaire (CRQ), and (3) the cost per 100 ml FEV₁ gained (26). Preferences were elicited by standard gambles, using the Maastricht Utility Measurement Questionnaire (MUMQ) (29), an adapted Dutch translation of the Health Utility Index (30) (Figure 1). The interviewer-administered version of the CRQ (31) and an MCID of 0.5 in domain score were used (32). The incremental number of subjects with an MCID was calculated, which is the inverse of the number needed to treat (NNT) (33).

View larger version (27K): [in this window] [in a new window]   Figure 1.   Health dimensions used to describe the subjects' own health state and the definition of the reference health state from the MUMQ.

Analysis

An "intent to treat" analysis was performed, including subjects with at least one follow-up assessment. Repeated measurement analyses were performed using the PROC MIXED procedure by SAS. For FEV₁ we assumed a nonlinear model (6) with baseline FEV₁, baseline PC₂₀, age, sex, height, allergy, smoking history, reversibility, and symptoms as covariates. QALYs were calculated as the area under the curve from the utility × time plots for each individual subject. Mean differences in QALYs were tested using t test statistics, corrected for individual baseline utility. Base-case cost effectiveness analyses were performed according to the Washington panel guidelines (26). Indirect cost was excluded from the base-case analyses. Early intervention was possible only after active detection. Therefore, a base-case and a secondary cost effectiveness analysis was performed. The first analysis was restricted to early medical intervention. The second analysis included the cost incurred by the detection phase of the program (19), that is, the cost per detected subject was added to the incremental cost of intervention.

Two subgroups were defined. (1) Subjects with reversible airway obstruction, defined as an FEV₁ reversibility,  10% of the predicted value after 800 µg salbutamol. (2) Current or ex-smokers who did not meet the reversibility criterion. Subgroup analyses were performed on postbronchodilator FEV₁, QALYs, and PC₂₀.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The screening identified 604 subjects (52.3% out of 1,155 subjects screened) with signs or symptoms of obstructive airway disease. A total of 384 subjects were willing to participate in the second stage of the detection program with quarterly monitoring. There were no significant differences in lung function, reversibility, age, and smoking history between the subjects who agreed to participate and those who refused (19). During the monitoring stage, 48 subjects met the criteria for persistent obstruction. This represented approximately 6.8% undiagnosed subjects in the general population (19). Six subjects met the criteria for bronchial hyperresponsiveness (approximately 0.9% of the general population) while 79 subjects met the criteria for an accelerated decline in lung function (approximately 12.5% of the general population). In total, 133 subjects were selected for the study: 82 of them (62%) gave informed consent and were randomized. Later, 8 subjects who had given informed consent had second thoughts and decided not to start using the medication; they dropped out after the baseline assessment (7 fluticasone propionate and 1 placebo). The remaining 74 subjects completed at least one follow-up assessment and were included in the intention-to-treat analysis. Seventy subjects (95%) attended the 1 yr follow-up assessment (2 fluticasone propionate and 2 placebo dropped out). The most frequently reported reasons for nonparticipation or dropping out were a general dislike of using medication and a perceived absence of (respiratory) complaints (34). There were no significant differences in subject characteristics between the participants (n = 74) and nonparticipants (n = 59). The subject characteristics are shown in Table 1. There were no significant differences in baseline characteristics between the subjects allocated to fluticasone propionate 250 µg twice a day (n = 33) and those allocated to placebo twice a day (n = 41).

Clinical Effects

The course of the post- and prebronchodilator FEV₁ is shown in Figure 2. Multivariate repeated measurement analysis revealed that baseline FEV₁, the level of reversibility, and the number of pack years were significant covariates. The multivariate model estimated an effect of 98 ml in postbronchodilator FEV₁ and 105 ml in prebronchodilator FEV₁ in favor of fluticasone propionate (p < 0.01 and p = 0.01, respectively). Analysis of the log-transformed PC₂₀ showed an effect of approximately half a dose-step in favor of fluticasone propionate, but this effect was not significant (mean difference: 0.49 dose-steps; p = 0.123). Nine exacerbation episodes were recorded in the fluticasone propionate group versus 13 in the placebo group. The percentage of symptom-free weeks throughout follow-up was not significantly different between the two groups (85.7% versus 85.2%; p = 0.92). The number of episode-free weeks was not significantly different between the two groups (96.8% versus 96.4%; p = 0.73).

View larger version (17K): [in this window] [in a new window]   Figure 2.   The course of the pre- and postbronchodilator FEV1 (SE bars). Estimated effect: 98 ml postbronchodilator FEV1 (p < 0.01); estimated effect: 105 ml prebronchodilator FEV1 (p = 0.01).

Cost Effectiveness Analysis

The mean annual cost of experimental medication was US$375 per subject (see Table 2). Early fluticasone intervention was consistently associated with lower health care cost and indirect cost, but these reductions were not significant. The cost of all other respiratory medication during follow-up was less than half than in the placebo group (US$15 versus US$33; p = 0.17). The average number of prescriptions per year for respiratory medication was 0.96 in the fluticasone group and 2.02 in the placebo group (p = 0.12). The average cost for GP consultations was low in the two groups. There were no referrals to specialists or hospital admissions. The average number of days that a subject had been unable to go to work or perform his or her normal daily activities due to respiratory symptoms was 3.2 in the fluticasone group and 11.0 in the placebo group: the associated indirect cost was US$225 and US$762, respectively. The distribution of indirect cost was very skewed and nonparametric testing did not result in significance (p = 0.93).

In the base-case cost effectiveness analysis, the incremental cost of early fluticasone intervention was US$35,100 per 100 treated subjects (Table 3). Inclusion of indirect cost resulted in net savings of US$18,600 per 100 treated subjects per year due to early fluticasone intervention. The cost associated with early detection by means of the two-stage detection program that preceded the trial was US$56,400 per 100 subjects (19); added to the incremental cost of early fluticasone intervention, the incremental cost of early detection and intervention was US$91,500 per 100 treated subjects (Table 3).

There were no significant differences in baseline utilities between a subject's own health state and the hypothetical health state (Table 1). Dominance violations (i.e., the description of the hypothetical health state was poorer than that of the subject's own health state, but the utility given was higher, or vice versa) did not occur. According to the interviewer, 93.4% of the standard gamble assessments had been well understood by the subjects. One subject refused to consider the standard gamble for religious reasons. A relatively large proportion (27.4%) of the health state descriptions collected during follow-up did not indicate any impairment on the six separate dimensions of generic quality of life (i.e., the health profile was 111111). The course of the mean utility of subjects' own health status is shown in Figure 3. After 6 mo, the mean difference was 0.0268 in favor of fluticasone propionate. At the end of the 1-yr follow-up, the mean difference was 0.0435 in favor of fluticasone propionate (Table 2). The estimated QALY gains were 2.70 QALYs per 100 treated subjects (p = 0.31, Table 3). The estimated mean difference in utility at 12 mo was 0.046 (repeated measurement analysis; p = 0.17). Assuming that this effect persists in the subsequent years, the QALYs gained by early fluticasone intervention would be 4.6 QALYs per 100 treated subjects per year. The baseline scores on the separate domains of the CRQ indicated mild impairment (Table 1). The development of disease-specific quality of life throughout follow-up did not show any significant differences between the fluticasone group and the placebo group with regard to the domains: emotions, fatigue, and mastery (Figure 4). The mean differences from baseline in these domains never exceeded the 0.5 threshold (MCID), indicating that on average, the differences found were not clinically relevant. The estimated effect of treatment from repeated measurement analysis at the end of follow-up was 0.26 (p = 0.21) for emotions, 0.26 (p = 0.33) for fatigue, and 0.07 (p = 0.66) for mastery. At the end of follow-up, the estimated mean difference in dyspnea between the two groups was 0.47 (p < 0.03). The mean within-subject difference in the fluticasone propionate-treated group (0.62; p < 0.01) indicated that on average, the improvements in dyspnea exceeded the 0.5 threshold. The proportion of subjects with an MCID in dyspnea was higher in the fluticasone group than in the placebo group (34% versus 13%; p < 0.04). The incremental number of subjects needed to treat or NNT was 4.76. 

View larger version (17K): [in this window] [in a new window]   Figure 3.   Preference based valuations of subjects' own health status: standard gamble utilities (SE bars). Estimated mean difference at end of follow-up: 0.046 (p = 0.17).

View larger version (15K): [in this window] [in a new window]   Figure 4.   Disease-specific health-related quality of life: mean differences per domain of the Chronic Respiratory Questionnaire (SE bars). Overall effect during follow-up: dyspnea (p = 0.09), emotions (p = 0.10), fatigue (p = 0.69), and mastery (p = 0.97). Estimated difference at the end of follow-up: dyspnea 0.47 (p < 0.03), emotions 0.26 (p = 0.21), fatigue 0.26 (p = 0.33), and mastery 0.08 (p = 0.67).

The cost effectiveness of early fluticasone treatment in the base-case scenario was US$13,016 per QALY (Table 3). The cost effectiveness of early detection and subsequent early treatment was US$33,921 per QALY. Inclusion of indirect cost resulted in dominance of fluticasone over placebo treatment and in a cost effectiveness ratio of US$14,031 per QALY for early detection and subsequent treatment. The cost for achieving an MCID in dyspnea was calculated as the product of the number needed to treat (4.76) and the mean incremental cost per subject; this resulted in US$1,674 per subject for early fluticasone intervention and US$4,361 per subject for early detection and treatment. The cost per 100 ml FEV₁ gained was US$359. Assuming a persistent effect equal to the utility gained at the end of follow-up, the cost per QALY of treatment in a subsequent year would decrease to US$7,630. A projection of the cost and effects in the long term, assuming fixed cost and effects, would therefore result in marginally improving cost effectiveness for the combination of detection and treatment (Figure 5).

View larger version (8K): [in this window] [in a new window]   Figure 5.   Cost-effectiveness of early detection and treatment over time, assuming fixed cost and effects. Cost and effects were discounted at a 4% rate.

Subgroup Analysis

Thirty-six subjects were labeled "reversible airway obstruction" while 24 were labeled "fixed airway obstruction" on the basis of repeated reversibility tests during the detection stage of the program. The remaining 14 subjects showed no signs of reversibility throughout the detection phase and had never smoked. The subject characteristics at the start of fluticasone intervention of the reversible and fixed subgroups are shown in Table 4. Analysis of the post bronchodilator FEV₁ showed that the course of the FEV₁ in the two subgroups was fairly similar in shape, both favoring fluticasone propionate. The estimated effect on post bronchodilator FEV₁ was 110 ml in the reversible subgroup and 99 ml in the fixed subgroup. The course of the mean utility in the two subgroups was fairly similar to that in the entire group: the estimated effect of fluticasone intervention was 2.6 QALYs per 100 treated subjects (reversible) and 3.3 per 100 treated subjects (fixed). Bronchial hyperresponsiveness improved significantly in the fluticasone group with reversible airway obstruction: approximately 1.4 dose-steps at 12 mo (estimated effect, corrected for baseline PC₂₀, 1.1 dose-steps; p < 0.04) (see Figure 6). No improvement in hyperresponsiveness was observed in the fluticasone group with fixed airways obstruction: approximately 0.4 dose-steps in favor of placebo (p = 0.86).

View larger version (13K): [in this window] [in a new window]   Figure 6.   Bronchial hyperresponsiveness in subjects with reversible airway obstruction (SE bars). Difference at t = 12 months: 1.4 dose steps, estimated effect, corrected for baseline PC20: 1.13 dose steps; p < 0.04.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study revealed that early fluticasone intervention resulted in a significant improvement in lung function and health-related quality of life and that early medical intervention may be worthwhile from a health economic perspective. Although it was expected that there would be little scope for improvement owing to the relatively good quality of life at baseline, clinically relevant improvements were observed.

In an earlier study we reported that the detection phase of the DIMCA study identified approximately 20% of the Dutch adult population with persistent airway obstruction, bronchial hyperresponsiveness, an accelerated decline in lung function, or a combination of these signs of obstructive airway disease (19). None of the subjects had been diagnosed prior to screening. These results are in line with those from foreign studies that indicated a higher prevalence of obstructive airway disease than was previously recognized (35). Underdiagnosis was primarily due to the fact that most subjects had not consulted their GP, despite the presence of persistent symptoms (38). In contrast with the normal diagnostic process, in which symptoms are presented to a GP, active detection enabled medical intervention for obstructive airway disease at an earlier stage than usual.

The effects of inhaled corticosteroids on the pre- and postbronchodilator FEV₁ in our actively detected, previously undiagnosed, and never before treated subjects were largely similar to those reported in patients whose obstructive airway disease was already diagnosed by their physician. During the first 3 to 6 mo, lung function improved, followed by a decline, approximately parallel with the decline observed in placebo patients. This "inverted hockey-stick" course of the FEV₁ has been reported in several previous studies, especially in patients with COPD (6, 8, 9). In patients with asthma, this is less clear, as there are indications that the decline in lung function after 6 mo may be somewhat improved in patients with asthma using inhaled corticosteroids when compared with patients not using inhaled corticosteroids (6). The "inverted hockey-stick" can probably be explained by the fact that in patients with COPD there is some inflammation that can be successfully treated by inhaled corticosteroids, but this cannot change the actual progression of the disease.

At the start of our screening program, no distinction was made between asthma and COPD, because it was our intention to select subjects in an early stage of their obstructive airway disease. In these early stages it is difficult to make the correct diagnosis, because patients' symptoms are not so pronounced that they seek medical care. After the screening we used time-related selection criteria to detect the patients that might benefit from fluticasone treatment. However, at the end of the study we were able to make a distinction between a subgroup of patients with a reversible obstruction (asthma) and a subgroup of patients with a fixed obstruction (COPD). This distinction was based on the current Dutch guidelines for GPs (39), which consider reversibility of the obstruction, after administration of a bronchodilator, as the most important criterion for distinguishing between asthma and COPD. Subgroup analyses were done on FEV₁, QALYs, hyperresponsiveness, but not on costs, because the variation in cost was too large to allow statistical tests in such small subgroups. The subgroup analyses showed that the effects of fluticasone on the main outcome parameters, FEV₁ and QALYs, did not differ significantly between those with reversible obstruction and those with a fixed obstruction. However, there was a difference in effect on bronchial hyperresponsiveness. A significant fluticasone treatment effect was observed in the reversible subgroup, but no treatment effect was observed in the fixed obstruction subgroup. This was expected, because in asthma, bronchial hyperresponsiveness is an accurate marker for inflammation (40) and oral or inhaled corticosteroid treatments have a well-documented antiinflammatory effect in asthma. In asthma corticosteroid treatment is able to reduce eosinophil counts in induced sputum (41, 42). This is in contrast with COPD, in which neither inhaled nor oral corticosteroids significantly affect neutrophil counts, granule proteins, or inflammatory cytokines in induced sputum (41, 43). In COPD, the increased bronchial hyperresponsiveness may not reflect inflammation, but only the degree of existing airway obstruction (44, 45).

The long-term effect of early fluticasone intervention on lung function was not addressed in this particular cost effectiveness study. This trial formed part of the DIMCA program, in which all screened subjects are followed up for approximately 10 yr after initial screening. This enables us to monitor long-term effects of early detection and intervention. The trial duration has been extended by 19 mo for subjects selected on the basis of an accelerated decline in lung function. The last 7 mo, all these subjects have been receiving fluticasone propionate 250 µg twice a day in a parallel-group, open-label study. The aim is to demonstrate whether loss of lung function is reversible after a 2-yr delay in treatment. The results of these long-term studies are being processed.

Treating asthma patients with fluticasone propionate resulted in improvements in health-related quality of life (46- 49), while the ISOLDE study showed relevant improvements in quality of life in patients with moderate to severe COPD (9, 12). In our study, there was only mild impairment in the subjects' quality of life at baseline. This left only small scope for improvement, but a clinically relevant and significant improvement in the dyspnea domain of the CRQ was observed. It can be postulated that this particular domain is sensitive to treatment, while the other domains are not. Fatigue, for example, is not very specific for obstructive airway disease in general: it can be due to poor physical fitness or comorbid conditions. On the other hand, given the relatively mild characterization, fatigue may not yet be a predominant symptom of the disease. This also holds true for emotions and mastery. The selected subjects were undiagnosed, thus the term "patient" is inappropriate. It was therefore unlikely that emotional problems and worries about losing control of the disease had any substantial impact in this particular group. The slow progressive nature of COPD may result in adaptation of life-style and in turn may explain why the subjects perceived little impairment in this respect. Inhaled corticosteroids may improve respiratory function and alleviate any impairments that the subjects become aware of after diagnosis, a phenomenon known as response-shift. In this way, there may be significant improvement, despite the relatively favorable situation at baseline.

Besides improvements in disease-specific quality of life, there was relevant improvement in subjects' preferences or utility. A frequent subject for criticism with respect to standard gamble utilities is their insensitivity to small changes in a person's health state. Although the difference in utility between the fluticasone and placebo groups was not significant, the course of the mean utility was in line with the other results. This may suggest that the improvement in utility was not merely due to chance, but that the study had insufficient power to detect any relevant improvement. When we set up this study, power calculations were primarily based on anticipated improvement in lung function. At that time, in 1991, no data were available on utility measurements in patients with obstructive airway disease, and, in particular, on the variance. Therefore, no power calculations could be made based on detecting relevant differences in mean utility.

Due to the acquisition costs of fluticasone, the annual direct medical costs per patient were US$351 higher in the fluticasone group than in the placebo group. The savings in costs of concomitant respiratory medication and the costs of GP consultations did not outweigh these acquisition costs. We have also found a reduction in indirect costs, which was far from significant due to the large variation in these costs.

In our study, the cost effectiveness of early medical intervention was expressed in cost per QALY. This expression allows direct comparison with other health care interventions, but, as opposed to cost-benefit analyses, it does not provide absolute or definitive decision rules. Despite the exponential growth in the number of published health economic evaluations, cost effectiveness analyses, performed in accordance with the recommendations of the Washington panel, are still scarce in the field of respiratory disease. Therefore it is premature to say that early medical intervention is a more efficient alternative than other interventions in COPD or asthma. In contrast with smoking cessation interventions, early medical intervention is relatively inefficient: smoking cessation strategies are estimated to cost somewhere between UK£212 and UK£873 per discounted life year saved (50), which is far less than early medical intervention. However, these results do not pertain exclusively to patients with COPD or asthma and one should bear in mind that a substantial proportion of people are not willing or able to give up smoking. Therefore, smoking cessation and early introduction of inhaled corticosteroids are not always alternative options. As far as we know, two cost-utility studies (i.e., cost per QALY analyses) have been performed on respiratory disease. In the Canadian study (51), ciprofloxacin treatment for acute exacerbations of chronic bronchitis was compared to the usual antibiotic treatment. A shift toward ciprofloxacin as first line treatment was associated with Can$18,588 per QALY, which was approximately as efficient as early intervention in our study. The other cost effectiveness study was performed on a Dutch lung transplantation program; there were significant gains in survival and quality of life (52). The associated cost effectiveness of transplantation varied from US$65,000 per QALY (improved supply scenario) and US$69,000 per QALY (restricted patient inflow scenario) to US$80,000 per QALY (baseline scenario). Although simple comparisons are prone to misinterpretations, these data do provide some sense of "value for money" in a broad spectrum of interventions for respiratory disease.

Calculating the cost per additional patient with a clinically relevant improvement (MCID) in dyspnea is an alternative for the cost per QALY ratio. The number of subjects that needed to be treated to improve one subject was 4.76 for dyspnea, whereas the associated cost was US$1,674. In a respiratory rehabilitation study (33) the number of patients with COPD who needed to be treated was 4.1 for dyspnea, whereas the associated costs were Can$47,548, approximately 20 times more expensive than early fluticasone intervention. Although the impact of rehabilitation on all four domains of the CRQ was more pronounced than that of early fluticasone intervention, the latter and above results indicate that the relative efficiency of COPD interventions is substantially higher in preventive interventions. This observation is in line with the results from preventive interventions in general: the cost per life year saved by prevention increased from a mean of US$5,000 for primary prevention to a mean of US$22,000 for tertiary prevention (53).

In conclusion, early treatment with fluticasone propionate 250 µg twice a day in subjects with objective signs of obstructive airway disease resulted in increased medication costs, but also in significant and relevant improvements in lung function and quality of life. Although no unambiguous distinction could be made between asthma and COPD in retrospect, there is evidence to suggest that the 1-yr effect of early medical treatment on the primary outcome parameters was approximately the same for asthma and COPD. Subgroup analysis on subjects with reversible obstruction and subjects with fixed obstruction revealed an effect on bronchial hyperresponsiveness in the former and no effect in the latter. Early fluticasone intervention appeared to be relatively cost effective, but this intervention was possible only after an extensive two-stage detection program. The cost effectiveness of early detection and medical intervention was much less efficient. However, the cost effectiveness of early detection and intervention marginally improves in subsequent years, provided that the effect on QALYs at the end of follow-up persists.