INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Recent studies have shown an increase in the number of cases of chronic obstructive pulmonary disease (COPD) and asthma in recent decades (1). Undoubtedly this increase will have an effect on the consequential costs. Given current cutbacks in health-care expenditure, it follows that there is an increasing demand for cost-effective interventions, but until now, health-economic analyses in COPD and asthma have been few.

Those health-economic analyses that have been done in COPD and asthma have focused on new or existing interventions, but have not considered the possible economic advantages and health gains of secondary prevention. It has been suggested that a large part of the economic costs of asthma are due to uncontrolled disease, and that proper treatment and monitoring of patients will result in considerable health gains (4). It is likely, however, that an increased effort to manage asthma and COPD more effectively will not result in a reduction of costs in the short run. Research has shown that substitution of more appropriate prophylactic medication for symptomatic medication in asthmatic individuals did not result in net savings (5, 6). However, these extra costs in the short term should be weighed against the health gains and possible cost reductions in managing COPD and asthma over the long term.

The advantages of reducing the prevalence of uncontrolled disease, both economically and physically, might very well apply to patients who are at an early stage of the disease, even before a confirmed diagnosis can be made. It has been shown that adequate treatment of newly detected asthma is effective (7), and that postponement of therapy can lead to irreversible loss of lung function (8, 9). However, patients at an early stage of the disease are either unaware of their condition or reluctant to consult their physician for respiratory symptoms (10).

The foregoing considerations resulted in establishment of the Detection, Intervention and Monitoring of COPD and Asthma (DIMCA) program in 1991. The program was established to answer the following questions: (1) What proportion of the general population has undiagnosed (and hence uncontrolled) asthma or COPD? (2) What proportion of the general population can be considered to be at an early stage of asthma or COPD? (3) What are the costs involved in the detection of these subjects? The common denominator of the two groups of subjects described here is that they are not currently scheduled in a program of treatment and follow-up. In the DIMCA program, the detection of these subjects was done in two consecutive stages: a screening of the general population and a 2-yr monitoring of subjects with positive findings on screening.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Design

The DIMCA program can be characterized as a randomized, controlled prospective study. The study was set up to detect undiagnosed cases of COPD and asthma but, especially to detect subjects with objective signs of COPD or asthma at the earliest possible stage. A random sample of registered subjects from 10 general practices, located in the eastern part of The Netherlands, were divided into two groups: an experimental or DIMCA group (n = 1,988) and a control group (n = 535). Because of the specific nature of the Dutch health-care system, virtually all Dutch inhabitants are automatically registered at a general practice. Therefore, this sample group can be regarded as a random sample of the general population. The experimental group was invited to participate in the first stage of the detection program: a single screening. Subjects with a positive screening result were then asked to participate in a second stage of the program: a 2-yr monitoring period. The control group received no intervention whatsoever and was kept unaware of the study (through randomized consent, according to Zelen [11]). This was done to ensure that the control group remained unbiased and could serve as a representative sample of patients seen in average daily practice (Figure 1). The study protocol was approved by the Ethics Committee of the University of Nijmegen Lung Centre. The results reported here pertain to the two-stage detection program with the experimental or DIMCA group (n = 1,988) only.

View larger version (22K): [in this window] [in a new window]   Figure 1.   Design of the DIMCA program. DIMCA 1 = subjects with persistently decreased lung function or increased bronchial hyperresponsiveness (BHR); DIMCA 2 = subjects with rapid decline in lung function with signs of BHR; DIMCA 3 = subjects with moderately increased decline in lung function or signs of BHR; false positives = subjects without objective lung function abnormalities throughout the monitoring period despite a positive screening result.

Subjects

All subjects were between 25 and 70 yr of age. All patients with known (i.e., diagnosed) asthma and COPD and those with congestive heart failure or lung diseases other than COPD or asthma, serious morbidity with a life expectancy of less than 4 yr, a severe physical or mental handicap, or corticosteroid dependency (oral or inhaled) were excluded. The remainder of the sample group was invited to participate in the first, screening stage of the program. Screening consisted of lung function measurements (see the section on LUNG FUNCTION, REVERSIBILITY, AND BHR) and of a symptom checklist, based on a modified Dutch version of the British Medical Research Council (MRC) respiratory questionnaire (12). Symptoms were categorized as either mild or severe (see the section on SYMPTOMS, SMOKING, AND ATOPY). Subjects were given a positive screening result if they met one of the following screening criteria:

1. Severe asthma- or COPD-related symptoms;
2. Bronchial obstruction defined as FEV₁/VC  predicted minus 1.64 SD;
3. FEV₁ reversibility  15% predicted at 15 min after administration of 800 µg salbutamol;

or at least two of the following three criteria:

1. FEV₁/VC  predicted minus 1 SD;
2. FEV₁ reversibility  10% predicted at 15 min after administration of 800 µg salbutamol;
3. Mild asthma- or COPD-related symptoms.

All subjects with a positive screening result (e.g., showing symptoms or objective signs) were then invited to participate in the second, 2-yr monitoring stage of the program. Monitoring consisted of periodic assessments of lung function, FEV₁ reversibility, symptoms, and costs at the general practice location (every 3 mo). Additionally, morning and evening measurements of peak expiratory flow rate (PEFR), medication use and current symptoms were recorded on patient diary cards during the first 3 mo. After the first 3 mo participants were asked to continue this recording once every week for the rest of the monitoring period. At baseline and after 12 and 24 mo, participants were invited to participate in extensive assessments at the university lung function laboratory. At the laboratory, the participants were subjected to spirometry (static and dynamic) and a histamine provocation test, and were asked to complete generic and disease-specific quality-of-life questionnaires. During the monitoring stage, the participants were treated on request by their general practitioner (GP), according to the Dutch Society for General Practitioners (NHG) standards.

During this monitoring stage, three mutually exclusive groups were distinguished. The first group consisted of subjects with persistently reduced lung function or increased BHR. They were detected during the first 6 mo of monitoring (hereafter called DIMCA 1). The detection criteria read as follows:

DIMCA 1. (1) During the first 6 mo of monitoring, at least two of four FEV₁ measurements equal to the predicted value minus 2 SD or lower; or (2) a PC₂₀ < 2 mg histamine/ml and FEV₁ reversibility  15%.

After completion of 1 yr in the monitoring phase, subjects with a rapid decline in lung function with signs of BHR were assigned to the second group (DIMCA 2) according to the following criteria:

DIMCA 2. (1) A decline in FEV₁ of more than 80 ml/yr during the first year of monitoring; and (2) a PC₂₀  8 mg histamine/ml and/or FEV₁ reversibility  10%.

After completion of 2 yr in the monitoring stage, subjects with a moderate increase in the decline in lung function or signs of BHR were assigned to the third group (DIMCA 3) as follows:

DIMCA 3 (moderate increase in decline in lung function). (1) A decline in FEV₁ of more than 40 ml/yr during the 2 yr of monitoring. Subjects with a decline of 40 to 80 ml/yr during these 2 yr then had to meet the additional criteria of: (2) A chronic productive cough for at least 3 mo/yr; or (3) at least one exacerbation per year.

DIMCA 3 (signs of BHR). Subjects had to meet at least two of the following three criteria: (1) Two PC₂₀ measurements  8 mg histamine/ml; (2) maximal FEV₁ reversibility  10%; or (3) at least one exacerbation per year.

Subjects who completed the two stages of the detection program but did not meet the foregoing criteria (DIMCA 1 to DIMCA 3) were regarded as "healthy." In retrospect, they were wrongly given a positive screening result; this group is referred to as "false positives" (Figure 1).

Symptoms, Smoking, and Atopy

Subjects completed a symptom checklist based on the modified Dutch version of the MRC questionnaire (12). A symptom score equal to the number of symptoms (range 0-8) was calculated. Wheezing, breathlessness at rest, coughing on most days for at least 3 mo, and an asthma attack or "allergic dyspnea" caused by specific and nonspecific triggers during the preceding 12 mo were considered severe symptoms (see the section on screening criteria).

Weather-induced coughing, phlegm production, or dyspnea; chronic phlegm production; and more than one period of coughing and phlegm production during the previous 2 yr were considered mild symptoms. Smoking status was assessed during the screening. Subjects were asked whether they were currently smoking and, if not, whether they had ever smoked. Smoking history was assessed in terms of pack-yr. Subjects were asked whether they had experienced eczema during childhood, as an indication of the presence of atopy.

Lung Function, Reversibility, and BHR

Spirometry (screening and annual assessments at the laboratory) was done with a Microspiro HI-298 (Chest Corp., Japan) (13). The quarterly spirometry done at the general practice office was performed with a hand-held spirometer (Microplus; SensorMedics, Walnut Creek, CA). After instruction, subjects were asked to perform at least three forced expiratory maneuvers from maximal inspiration. The FEV₁ corresponding to the maneuver with the highest sum of the FEV₁ and FVC was recorded. Predicted values were calculated according to European Respiratory Society (ERS) guidelines. The degree of reversibility was defined as the change in FEV₁ at 15 min after inhalation of 800 µg salbutamol relative to the predicted value (in percent). Fifteen minutes after administration of salbutamol, ipratropium bromide (160 µg) was administered. Maximal reversibility was defined as the change in FEV₁ at 45 min after inhalation of 160 µg ipratropium bromide relative to the predicted value (in percent). During each subject's annual visit to the laboratory, a histamine provocation test was performed (in accordance with the protocol of Cockcroft and associates [14]). The provocative concentration was estimated by means of linear interpolation, and was subsequently ²log transformed. This transformation resulted in continuous values from 6 (20% decline after inhalation of a saline solution) to +5 (32 mg/ml histamine).

Exacerbations

An exacerbation was defined as a situation in which at least two of the following three criteria were met: (1) an episode with increased (productive) coughing and/or dyspnea and/or wheezing; (2) change in sputum color; and (3) (increased) use of bronchodilator drugs. These events were registered by the subjects themselves in the diary and on the weekly cards, and on consultation forms filled in by the physician in attendance.

Measurement of Costs in Three Scenarios

The following cost categories were included in the calculation of costs of the detection program: (1) costs of spirometry and assessment of FEV₁ reversibility and BHR (histamine provocation test); (2) costs of administration and organization, including the costs involved in summoning all subjects, sending reminders, and scheduling appointments; (3) time costs to the participants, including traveling time; and (4) transportation costs.

Cost data were collected in natural units and subsequently multiplied by the cost per unit. The costs per unit of spirometry and assessment of FEV₁ reversibility and BHR were estimated by means of 1996 reimbursement fees (source: 1996 Centraal Orgaan Tarieven Gezondheidszorg [COTG] data). To avoid inequalities between wage earners and participants without paid work, all time costs were valued equally. Time costs were valued according to the net hourly wages of the working participants. Transportation costs were valued in accordance with current guidelines for economic analyses in The Netherlands (15). The costs for administration and organization were valued as being equal to the reimbursement fees for administration and organization in a current mass-screening program for cervical cancer in The Netherlands (source: 1996 COTG data). The costs for the three following detection scenarios were calculated: (1) Screening followed by a 6-mo monitoring period. During this period only subjects meeting the DIMCA 1 criteria could be identified. (2) Screening followed by a 12-mo monitoring period. During this period subjects meeting either the DIMCA 1 or the DIMCA 2 criteria could be identified. (3) Screening followed by a 24-mo monitoring period. During this period subjects meeting either the DIMCA 1, the DIMCA 2, or the DIMCA 3 criteria could be identified (full program).

The costs involved for false positives were included in all three scenarios since these costs were unavoidable. False positive results could be identified in retrospect only. The costs were converted to United States dollars on the basis of purchasing power (1 United States dollar = 2.08 Netherlands guilders).

Sensitivity Analysis

In a univariate elasticity analysis, the key parameters were varied individually with a constant percentage (an arbitrary 10%) to assess the impact of variation on the outcome. Outcome was defined as the total costs of the full program and the costs per detected case. A multivariate, semistochastic sensitivity analysis was then performed. In the multivariate analysis, the costs and the proportions of cases detected were varied simultaneously. Under "optimistic" assumptions, high proportions for all three detection groups were combined with low estimates for the total costs. Under "pessimistic" assumptions, low proportions were combined with high estimated costs. A high or low proportion or prevalence was defined as the observed proportion in the sample plus or minus 1 SEM. Low or high cost estimates were derived by a reduction or increase of the unit prices by a constant 25%.

Withdrawal, Refusal, and False Negatives

Data for a sample of subjects who refused to participate in the screening or who withdrew from the study during the monitoring phase were collected to check for possible biased recruitment or selection. The data consisted of FEV₁% predicted, FEV₁ reversibility, sex, age, atopy, BHR symptoms, and pack-years, when available. Special attention was paid to those subjects who refused to participate in the screening stage of the study. Information on pulmonary function and symptoms was considered particularly critical in these subjects. Therefore, a random sample of subjects who refused to participate in the original screening stage were given the opportunity to be screened at home, in their own time. This screening at home did not include an assessment of reversibility of FEV₁, since a reluctance to inhale medication was the predominant reason for refusal to participate in the first place. The home visits were made to check for possible biased recruitment only. To assess the test characteristics of the screening, and particularly the number of false negatives, a random sample of subjects with a negative screening result were asked to participate in a "reduced" monitoring program for 2 yr. As with those subjects who refused, only FEV₁ and symptoms were monitored, and the assessments took place at the subject's home. Spirometry and the spirometer used were identical to those in the detection program. The costs for these assessments were not taken into account.

Statistical Analysis

Differences in patient characteristics were tested by means of independent-sample t tests (normally distributed, continuous variables). Continuous variables with skewed distributions were tested by means of Mann-Whitney U tests. Dichotomized variables were tested by means of chi-square tests. The decline in FEV₁ during the monitoring stage of the study was estimated by means of linear regression. The proportions of patients detected during the program (e.g., DIMCA 1 to DIMCA 3) were extrapolated to the entire general population. To account for dropout and selection at different points during the program, a proportional hazards technique was used for this extrapolation (Life Table procedure, SPSS 6.1 for Windows; SPSS, Inc., Chicago, IL).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Of the initial random sample (n = 1,988), 239 subjects failed to meet the inclusion criteria for the study. Only 29 of the excluded subjects had a confirmed diagnosis of COPD or asthma and were scheduled in a treatment plan (i.e., were considered to have controlled disease) at the time. The remaining 1,749 subjects were invited to participate in the first screening stage of the study. Of these 1,749 subjects, 1,155 agreed to participate. Of those who refused to participate a random sample of 34 subjects was taken, all of whom agreed to be screened at home. This sample was used to check for possible biased recruitment. Those subjects who refused to participate were no different from those who participated except for their symptom score (Table 1).

The first stage of the study resulted in the identification of 604 (52.3%) subjects with symptoms or signs of COPD or asthma. They met the screening criteria for the study, were labeled as positive for COPD or asthma, and were invited to participate in the second, 2-yr monitoring stage of the study. A total of 384 subjects agreed to participate. The remaining 220 subjects either refused or were not able to participate for logistical reasons. These subjects were comparable to those who participated except for a relatively small difference in symptom score (Table 1). During the 2-yr monitoring period, 54 subjects withdrew from the study. Those who withdrew were comparable to those who remained with respect to the screening parameters, but were younger on average and showed less BHR at baseline (Table 1).

During the first 6 mo of the monitoring stage of the study, 54 subjects met the DIMCA 1 criteria. Once identified, they were withdrawn from monitoring. At the end of the first 12 mo, the decline in lung function of the remaining subjects was assessed: 79 subjects met the DIMCA 2 criteria and were also withdrawn from monitoring. At the end of the 2-yr monitoring period, 119 subjects met the DIMCA 3 criteria. No objective signs of COPD or asthma were found in the remaining 78 subjects (as noted earlier, 54 subjects dropped out during these 2 yr), they were considered a posteriori to have false-positive diagnoses. These results were extrapolated to the entire undiagnosed general population. Approximately 7.7% of the general adult population showed persistently decreased lung function or increased BHR. Approximately 12.5% of the population showed a rapid decline in lung function with signs of BHR. Another 19.4% had a moderate increase in the decline in lung function or signs of BHR (Figure 2). Analysis of lung function in a random sample of subjects with a negative screening result (n = 79), showed that none of the subjects met the criterion of having persistently reduced lung function. In this respect, the predictive value of a negative screening result was 100%.

View larger version (22K): [in this window] [in a new window]   Figure 2.   Results of two-stage detection program, extrapolated to the undiagnosed general population. Negative = proportion of subjects in the general population with a negative screening result; DIMCA 1 = proportion of subjects in the general population with persistently decreased lung function or increased BHR; DIMCA 2 = proportion of subjects in the general population with a rapid decline in lung function with signs of BHR; DIMCA 3 = proportion of subjects in the general population with a moderately increased decline in lung function or signs of BHR; false positives = proportion of subjects without objective lung function abnormalities throughout the monitoring period despite a positive screening result. (Negative + DIMCA 1 + DIMCA 2 + DIMCA 3 + false positives = 100%).

The percentage of current smokers and ever-smokers, and the average number of pack-years in the entire screened sample are shown in Table 2. The screening results showed two groups that were significantly different with respect to smoking status, but the extent of the differences indicated that neither current smoking nor smoking history were sensitive and specific alternative screening parameters. This was confirmed by a stratified analysis of FEV₁ in current smokers, in which current smokers with a positive result showed a significantly reduced FEV₁% predicted (90.4%), whereas current smokers with a negative result had normal lung function (98.7%). There was no difference in smoking status between identified subjects with objective signs of COPD or asthma and the rest of the screened sample. The percentage of ever-smokers among those subjects who were identified as having COPD or asthma was similar to the percentage of ever-smokers in the remainder of the screened sample (59.9% versus 63.7%). A small but statistically significant difference in the number of pack-years was observed.

The costs per unit, number of units used, and total costs of the detection program are shown in Table 3. The total costs of screening were US$22,668. The total cost of monitoring was US$95,549. With both stages taken together, the cost of the two-stage detection program amounted to US$118,217, including the costs involved for false-positives. The average follow-up was approximately 18 mo (SE = 0.44).

Table 4 shows the costs per detected case in the three different scenarios. Detection of subjects with persistently decreased lung function or increased BHR (DIMCA 1) cost US$51,452, equivalent to US$953 per detected case. A 6-mo extension of the program cost an additional US$23,639. This made it possible to detect subjects with a rapid decline in lung funtion (DIMCA 2). Since the increase in the detection rate was more than proportional to the increase in cost, the cost per detected case decreased to US$564. The relative efficiency of this extension is expressed in the incremental costs per detected case (Table 4, Column 7). The extension of monitoring for another year cost an additional US$43,126 and made it possible to detect 119 subjects who met the DIMCA 3 criteria. This extension was approximately as efficient as the 6-mo extension, since the incremental cost per detected case was only slightly higher (US$363 versus US$318). The total cost for the entire two-stage program amounted to US$118,217, equivalent to an average of US$469 per detected case.

The elasticity analysis revealed that the outcome of the detection program in terms of cost was not particularly sensitive to changes in any of the cost parameters (Table 5). The outcome was most sensitive to changes in the cost of spirometry: a 10% change in the cost of spirometry would result in a 3.3% change in the total cost of the detection program. The multivariate analysis revealed that simultaneous variation of cost and outcome parameters resulted in a 35% to 37% reduction of the costs per detected case, depending on the scenario chosen (optimistic estimates). The pessimistic estimates of the costs per detected case were 48% to 53% higher than the costs found in the program.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The two-stage detection program revealed that a large proportion of the general undiagnosed population showed symptoms and objective signs of COPD and asthma. The first, screening stage of the program showed that approximately 50% of the general population had either respiratory symptoms or objective signs of obstructive airways disease. The consecutive second stage showed that approximately 20% of the general undiagnosed population met either the DIMCA 1 or the DIMCA 2 criteria. In the case of these subjects, treatment should be considered, since earlier studies have found it to be effective (7, 16). The cost involved in the detection of these subjects, considered to have uncontrolled disease, was US$564. Extension of monitoring for another 12 mo identified another 19.4% of the population who could be considered to be at a very early stage of or developing COPD or asthma. To date, it is unclear whether therapy would have a preventive effect on these subjects. The average cost involved in the additional detection of this experimental group was US$469.

Although the two-stage detection program was feasible in the setting in which it was established, the protocol may have practical limitations in other health-care environments such as the United States. In The Netherlands, a network of general practitioners exists, covering virtually the entire general population. This enables long-term follow-up of patients, but also follow-up of presumably healthy individuals. In our opinion, early detection of COPD and adult asthma is only possible through repeated measurements and evaluation over the course of time. Therefore, health-care environments should meet these requirements before secondary prevention, as advocated in the DIMCA program, is attempted. The results of the DIMCA study may trigger health-care organizations in other countries to (re)organize their services so that long-term monitoring is possible. On the other hand, current health plans provided by health maintenance organizations (HMOs) and regular examinations in the setting of occupational medicine may incorporate early detection protocols. Such protocols will not address the entire general population, but we believe that secondary prevention on a limited scale will also result in health gains and substantial future savings.

The efficacy of treatment of DIMCA patients is currently being assessed in a series of three randomized clinical trials comparing fluticasone at 250 µg twice daily with placebo. The preliminary results of treating DIMCA 1 patients showed statistically significant improvements in FEV₁ and PC₂₀ (17). This confirmed our assumption that considerable health gains can be achieved in the case of persons in whom COPD or asthma is as yet undiagnosed and untreated. The rationale behind early intervention is that irreversible loss of lung function may be prevented. It has been shown that the accelerated decline in lung function in COPD and asthma can be decreased, and that a delay in the initiation of medical therapy may cause irreversible loss of function (7, 16). Therefore, early medical intervention should perhaps be considered, although it is by no means an alternative to interventions directed toward smoking cessation. Smoking cessation is a goal of paramount importance, but is not easily attained. In reality, a substantial proportion of the population at risk for developing COPD or asthma is unwilling to consider giving up smoking, and unfortunately, smoking cessation programs do not always show success. On the other hand, not all obstructive airways disease is smoking-related. Considering these realities, early medical intervention may be justified as a therapeutic possibility additive to current preventive measures. The objective data on decline in lung function gathered during the monitoring stage of a detection program may well help physicians to motivate smokers to quit and to increase their adherence to cessation programs. However, a complete evaluation of health gains from and the cost effectiveness of early detection and intervention is not possible at present. The data from the three respective treatment experiments described in the present report will not become available in the immediate future, since patients are currently being treated for a period of 2 yr.

In the DIMCA program, no distinction between COPD and asthma was made at first. Instead, a number of objective criteria indicative of obstructive airways disease were formulated (DIMCA 1 to DIMCA 3). The purpose of defining sets of criteria instead of using a diagnosis was to focus primarily on detecting undiagnosed obstructive airways disease at an early stage. As a consequence, many subjects did not meet all the required criteria for a formal diagnosis. A second reason for using sets of criteria was that we anticipated that a substantial number of subjects would have signs of COPD and asthma simultaneously. A classification as either diagnosis would then be difficult if not impossible. The results of the detection program substantiated this belief. The great majority of subjects who met the DIMCA 1 criteria were detected solely on the basis of a persistently reduced FEV₁. This would suggest that these subjects primarily had COPD. However, these subjects had a PC₂₀ of approximately 2 mg/ml and were 44 yr of age on average (range: 26 to 69 yr). Together, these characteristics would make an unequivocal diagnosis of COPD difficult in the least. The fact that we initially made no distinction between COPD and asthma does not necessarily mean that we will not do so a posteriori. The analysis of treatment effects will be stratified on the basis of likely diagnosis.

Considerable attention was paid to smoking status in the screening and throughout the monitoring stages of the study. However, smoking status or smoking history was not part of the screening criteria. The main reason for this was that we intended to detect airways morbidity at an early stage, and not merely to identify a high-risk population. Although smoking is widely accepted as the predominant risk factor for COPD, not all smokers will develop COPD, and vice versa, not all nonsmokers will be safeguarded against it. Moreover, in most cases, the time between exposure to the risk factor and the onset of disease is long, and may well extend beyond the scope of a 2-yr monitoring period. The results showed that the presence of symptoms or signs of COPD or asthma during the screening stage of the study was associated with smoking, but this association was, perhaps surprisingly, weak. If the screening had included current smokers, a positive result would have been found for 170 current smokers with normal lung function. Perhaps even more remarkably, 60% of all detected subjects with objective signs of COPD or asthma were not currently smoking, and would have been missed if screening had been based on current smoking only. That the proportion of current and past smoking in identified subjects was not different from that in those who were not identified clearly indicated that smoking as a screening criterion was very unlikely to be sensitive or specific.

It is hard to say whether the costs per detected case found in this study are high or low. Table 6 shows the costs per detected case in several other prevention programs. The reader should bear in mind that the methods used and the costs appreciated in these studies may vary. No effort was made to correct these differences, nor was any correction made for inflation or differences in exchange rates. One should be very cautious, therefore, in drawing definite conclusions from Table 6; it should instead be interpreted as indicative. Taking these drawbacks into consideration, the costs involved in the detection of uncontrolled COPD or asthma (Scenario 2) compare well with those for the detection of other diseases. The costs may be considered comparable to those associated with the detection of abdominal aortic aneurysms; only the detection of hypertension in blue collar workers seems less costly. Again, it is stressed that the true cost-effectiveness of such programs can be evaluated only if the health gains are taken into consideration. A cost per quality-adjusted life year (QALY) analysis of the DIMCA program will be performed in due time.

The number of false-positive results in the present study was relatively high, but this had been anticipated in the design of the DIMCA program. Relatively mild screening criteria were purposely chosen to ensure that all subjects with signs and symptoms of possible morbidity would show a positive result. In retrospect, the two-stage detection program could be optimized by selecting only the more sensitive and specific parameters for detection of COPD and asthma. Implementation of an optimized program might result in a reduction in false-positive results and, consequently, in lower costs per detected case. For cases of persistently decreased lung function, the predicted value of a negative screening result was 100%. The other detection criteria could not be tested, since this would have required information on FEV₁ reversibility and/or BHR. However, there are indications that the proportion of false-negatives with the DIMCA program is very small. We followed the complete scope of postscreening health-care utilization by a random sample of 200 subjects with a negative screening result for an average period of 3.6 yr. The use of health care for respiratory reasons, and the costs involved, were very low in this sample and were mostly associated with common colds or influenza. None of the 200 subjects with a negative screening result had a diagnosis of COPD or asthma made in this period, nor did they receive any specific asthma or COPD medication. A significant difference in COPD- or asthma-related use of health care resources by subjects with a positive screening result and subjects with a negative result confirmed that the screening was successful. In addition, the analysis of resource utilization data revealed that the DIMCA program was not associated with additional use of health-care resources beyond those consumed in average daily practice (44).

The results of the detection program confirmed that underdiagnosis of COPD and asthma is a substantial problem. However, it is not justified to blame the general practitioner for this: 74% of all subjects with symptoms or signs of COPD or asthma in the present study never consulted their general practitioner for respiratory complaints (10). This high percentage was found regardless of the severity of symptoms or lung function impairment. Of the 1,749 invited subjects, 1,155 (66%) agreed to participate in the screening stage of the study. This could imply that screening on a large scale will not be attended by a third of the target population. In contrast to the efforts made during the monitoring stage of the study, no effort was made during the process of invitation to undergo screening to increase the participation rate, such as by sending reminders or personal communications. During the monitoring stage, the practice assistants organized the three monthly measurements themselves. Their motivation and personal communication with the participants undoubtedly contributed to the high attendance in the monitoring stage of the study. The main reasons for nonparticipation were absence of respiratory complaints, time constraints, and reluctance to take medication (in order to assess FEV₁ reversibility). Fifty-four percent of the nonparticipants responded that they would have participated had the detection program been part of routine medical care (45). With respect to possible biased recruitment or selection, the nonparticipants and dropouts were comparable in terms of FEV₁% predicted. The differences found with respect to symptoms and BHR might indicate that nonparticipants and dropouts were slightly healthier. In turn, this might imply that the number of cases found in the program could be somewhat overestimated. The available data for all dropouts were used in a logistic regression procedure to categorize subjects according to the three sets of study criteria. This categorization revealed that similar proportions of subjects could have been expected to meet the DIMCA criteria had they continued in the monitoring stage of the study.