INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Allergy manifested as increased levels of serum immunoglobulin (Ig)E, positive skin tests, or eosinophilia is frequently present in subjects with asthma (1). Eosinophilia has been shown to increase the risk for asthma mortality in patients with mild to severe asthma (2). There are indications that eosinophilia may be important not only in the development of asthma and its associated mortality, but also in the development of chronic obstructive pulmonary disease (COPD) (3). For example, in a general population sample, eosinophilia and positive skin tests were associated with low lung function (4), and positive skin tests were a risk factor for a more rapid decline in FEV₁ (5, 6). In addition, Mensinga and coworkers (7) found that eosinophilia was not only associated with asthma attacks, but was also associated with chronic cough, bronchitis episodes, and shortness of breath experienced by subjects walking with other persons of the same age on level ground, independent of positive skin tests.

Mortality in the general population has been found to be increased in subjects with low FEV₁ (8) and rapid decline in FEV₁ (9), as well as in smokers (10). Because allergy markers such as eosinophilia and positive skin tests are associated with low FEV₁, rapid decline in FEV₁, and presence of respiratory symptoms, we investigated whether these allergy markers predict COPD mortality in the general population. We also studied whether COPD mortality, both as the primary cause and combined primary and secondary causes of death, was increased if subjects had both allergy and either respiratory symptoms or a low lung function.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Population

Epidemiological field studies on risk factors for asthma and COPD were carried out in three Dutch communities. The population selection has been described (11, 12). In brief, in 1964 a survey was carried out in Meppel in which all men age 40 to 65 (2,724 subjects) were invited to participate and complete a questionnaire. All men whose answers indicated symptoms of asthma or COPD and a random sample of the remainder were subjected to the objective measurements described subsequently. In 1965 a questionnaire survey was carried out in Vlagtwedde, a rural area in the northeast part, and in Vlaardingen, an urban area in the western part of The Netherlands. In Vlagtwedde, all inhabitants age 40 to 64 (2,386 subjects) were invited to complete the questionnaire. Fifty percent of this group, randomly selected, were invited for elaborate investigation with objective measurements. In Vlaardingen, a random sample of all inhabitants in the General District Register of that age group were invited (1,512 subjects) for both the questionnaire and objective measurements. These surveys were followed by a survey of young people age 15 to 39 in 1967 in Vlagtwedde (2,449 subjects) and in 1969 in Vlaardingen (2,262 subjects) as part of a longitudinal study. In some cases data on eosinophils were not available in the first survey but were available in the first follow-up survey in 1970 in Vlagtwedde or in 1972 in Vlaardingen, in which case data from this first follow-up were used (773 cases). A total of 7,735 subjects had both a peripheral blood eosinophil count and skin tests available; 179 subjects were excluded because of missing data on respiratory symptoms, lung function, smoking habits, or body mass index (BMI). A total of 7,556 men and women had peripheral blood eosinophil counts, skin tests, and complete data for all other covariates.

Peripheral Blood Eosinophil Count

Peripheral blood eosinophil counts were assessed in a 1:11 dilution of peripheral blood with a Bürker counting chamber (7, 12). Eosinophilia was defined as  275 cells/mm³ of blood. This cutoff point was significantly associated with respiratory symptoms generally considered allergic in nature (persistent wheeze and asthma attacks) (7).

Allergen Skin Testing

Four common aeroallergens were applied intracutaneously to the volar forearm: house dust, mixed pollen, epidermal products, and mixed molds (Diephuis, Groningen, The Netherlands) (7, 12). Wheal diameters for each allergen were measured to the nearest half millimeter and coded on a six-point scale (0 = 0 to 5 mm, 1 = > 5 to 7.5 mm, 2 = > 7.5 to 10 mm, 3 = > 10 to 12.5 mm, 4 = > 12.5 to 15 mm, 5 = > 15 mm). Scores for the four allergens were added to a skin test sum score (minimum 0, maximum 20). Subjects had positive skin tests if the skin test sum score was  3. In a previous analysis the prevalence of positive skin tests by this definition was found to be optimally and significantly associated with respiratory symptoms generally considered allergic in nature (persistent wheeze and asthma attacks) (7). A histamine biphosphate solution was used as a positive control (12, 13). Eleven subjects had no data, and 1,155 subjects had no reaction to the positive control; these subjects were included in the analysis. The majority of these subjects were tested in 1964 (641 subjects) and 1965 (501 subjects), and were 40 to 64 yr of age.

Questionnaire

Data on age, sex, smoking habits, and respiratory symptoms were collected by means of a Dutch version of the British Medical Research Council's standard questionnaire (14, 15). Interviews were performed by trained interviewers. Smoking was considered as a categorical variable: ex-smokers had stopped smoking at least 1 mo before the examination, current smokers smoked  1 cigarette a day and were divided into subgroups of < 15 or  15 cigarettes a day. Pipe and cigar smokers were also considered as smokers, in such a way that 1 g of tobacco was considered as 1 cigarette. Asthma was considered to be present if an affirmative answer was given to the question whether a subject had ever experienced attacks of shortness of breath with wheezing at rest (asthma attacks). Wheeze was defined as a wheezing or whistling sound in the chest on most days or nights. Chronic cough and chronic phlegm were defined as cough or phlegm production on most days or nights for as much as 3 consecutive months each year during the winter. Dyspnea grade III or more was present if subjects reported that they were troubled by shortness of breath when walking with other persons of their own age on level ground.

Spirometry

Inspiratory vital capacity after a deep expiration and then FEV₁ were measured with a water-sealed spirometer (Lode Spirograph D53; Lode Instruments, Groningen, The Netherlands). Subjects performed the maneuver until two technically satisfactory tracings were produced, the higher value of these tracings being taken as the baseline measurement (4, 12). Percent predicted reference values were calculated with regression coefficients derived from analysis of all asymptomatic subjects regardless of their smoking habits who took part in 1965-1969 or 1970-1972 in the populations of Vlagtwedde and Vlaardingen. We computed for both sexes separately regression equations for FEV₁ as a function of age and height, with an age cutoff of 21 yr. The actual measured FEV₁ value was expressed as the percentage of the predicted FEV₁, calculated from these regression equations.

BMI

BMI was calculated as weight divided by height² and divided into 4 classes according to WHO criteria: underweight < 18.5, normal weight 18.5 to 25, overweight 25 to 30, and obesity  30 kg/m² (16).

Follow-up

Subjects were traced until March 10, 1995, and their vital status (5,135 alive, 2,315 dead, 106 lost-to follow-up) was assessed with 99% success. Survival time was calculated for each subject from the date of entry into the study (between 1964 and 1972) until the end of follow-up, either (1) March 10, 1995, for subjects registered at the municipalities as being alive, (2) the date of death for subjects identified in the death register of the municipalities, or (3) the last registration of subjects lost to follow-up, for example, date of last survey attended or date of move when the new address could not be traced. Causes of death were obtained from Statistics Netherlands in Voorburg which classified the death causes according to the International Classification of Disease (ICD). We used ICD codes for mortality from COPD of three versions of the ICD according to the classification of Mackenbach (17, 18): version 7 (code 501, 502, 526, 527.1), version 8 (code 490-492, 518), and version 9 (code 490-492, 494, 496). For mortality from asthma the ICD codes in these versions were in version 7 (code 241), in version 8 (code 493), and in version 9 (code 493).

Statistical Analyses

The associations between allergy markers and COPD mortality were estimated with the proportional hazards model of Cox (19). Time was defined from the initial examination until mortality from COPD as the endpoint of interest. Mortality from COPD was analyzed twice, once restricting to the primary or underlying death cause and once including all subjects with either a primary (maximum 1) or a secondary (also called contributory) death cause of COPD (maximum 3). Censoring took place when the subjects were still alive at March 10, 1995, were lost to follow-up (n = 106, 1%), or died of a cause other than COPD (20). A proportional hazards model accounts for varying intervals in follow-up between subjects, e.g., censoring, and permits control for potential confounding effects of other risk factors. We controlled for gender, age, smoking habits, and city associated with both allergy (13), and with low level of lung function (4) or with symptoms of COPD (7), and therefore potential confounders for COPD mortality. We also controlled for underweight, known to be associated with poorer survival in COPD (21). To test whether subjects with allergy markers and either respiratory symptoms or a low level of lung function had a higher risk of COPD mortality, an interaction term of one of the respiratory symptoms or lung function with the allergy markers was added, one at a time, to Model 1 and, if significant, presented in Model 2. Statistical significance was defined as a p value  0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Table 1 shows that 12% of our population (921 subjects) had eosinophilia and 13% (1,012 subjects) had positive skin tests at the start of the study. Subjects with eosinophilia had a lower level of lung function and were significantly more likely to live in Vlaardingen, have positive skin tests, dyspnea  III, wheeze, and a history of asthma attacks than subjects without eosinophilia. Subjects with positive skin tests were younger, had lower BMI, and were significantly more likely to live in Vlaardingen, have eosinophilia, a history of asthma attacks, and less cough than subjects without positive skin tests.

A very high percentage of all subjects were traced (99%) with a maximum of 30 yr of follow-up. In 1995, 5,135 of the 7,556 subjects were still alive, 106 lost to follow-up, 121 died of COPD as the primary death cause, and 2,194 subjects died of other primary causes of death of whom 137 had COPD as a secondary cause. The mean follow-up time was 24.0 ± 6.5 yr for all subjects. During follow-up 2,315 deaths (31%) occurred; death was more prevalent in those with eosinophilia (35%) and negative skin tests (33%). A total of 121 subjects (5%) died with COPD as the primary cause, and 258 subjects (11%) with COPD as a primary or secondary cause.

COPD As the Primary Cause of Death

Table 2 shows the results of the association of the allergy markers and COPD mortality (121 deaths) as the primary cause of death (left side of table), and with further adjustment for the interactions between the allergy markers and a history of asthma attacks (lower two lines). The latter model shows that subjects with both eosinophilia and a history of asthma attacks had increased risk of mortality from COPD as the primary death cause (relative risk [RR] = 1.06 × 0.82 × 5.52 = 4.80; confidence interval [CI] 1.92 to 11.93) compared with subjects without eosinophilia and a history of asthma attacks. Interactions of the allergy markers with other respiratory symptoms or lung function were not significant. Dyspnea  III and FEV₁ percentage of predicted < 80% were associated with increased COPD mortality, whereas the effect of having an FEV₁ of 80 to 100% predicted was borderline significant. The respiratory symptoms asthma attacks, wheeze, cough, and phlegm were not associated with increased COPD mortality. As expected, male gender, age, and smoking (only  15 cigarettes/day) were all associated with increased COPD mortality in both models. City did not contribute. Moreover, underweight (BMI < 18.5 kg/m²) contributed significantly to COPD mortality.

COPD As a Primary or Secondary Cause of Death

When combined primary and secondary death causes were investigated (right part of Table 2), positive skin tests were not significantly associated with an increased risk of COPD (RR = 1.31; 95% CI 0.86 to 1.99). The model with interaction terms (lower two lines of Table 2) shows that only subjects with both eosinophilia and a history of asthma attacks but not other respiratory symptoms or low lung function had increased COPD mortality (RR = 1.10 × 1.04 × 3.41 = 3.90; CI 2.05 to 7.40) compared with subjects without eosinophilia and asthma attacks. Dyspnea  III and both subgroups FEV₁ percentage of predicted < 80% and 80 to 100% were associated with increased COPD mortality. The respiratory symptoms cough and phlegm were not associated with increased COPD mortality, whereas the respiratory symptoms asthma (RR = 1.44; CI 0.99 to 2.09) and wheeze (RR = 1.39; CI 0.98 to 1.97) were borderline significantly associated with increased mortality from COPD. When adjusted for the interaction between eosinophilia and asthma attacks, the association between wheeze and COPD mortality became statistically significant. Furthermore, subjects with positive skin tests and an FEV₁ percentage of predicted 80 to 100% had a reduced risk of combined primary and secondary COPD mortality. In contrast to analyses restricted to the primary death cause, all smoking habits, i.e., former as well as current smoking < 15 cigarettes/day, were clearly associated with increased mortality of combined primary and secondary death causes of COPD.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study shows that in addition to well-known risk factors such as smoking and low lung function, eosinophilia in combination with a history of asthma attacks is also associated with increased risk of mortality from COPD both as the primary and combined primary and secondary causes. Eosinophilia in combination with either one of the other respiratory symptoms (cough, phlegm, dyspnea  III or wheeze) or combined with a decreased lung function is not associated with an increased mortality risk of COPD. Positive skin tests are not associated with mortality from COPD.

It is possible that "asthma attacks with eosinophilia" predicts an increased risk of mortality from COPD in addition to the above well-known risk factors for COPD mortality. We know that unstable or clinically active asthma is associated with sputum eosinophilia (22, 23) and peripheral blood eosinophilia (24, 25). More severe symptoms and more severe disease in general may be associated with more extensive airway remodeling (26). We can only speculate that this leads to irreversible airway obstruction (27) and COPD, and finally to mortality from COPD.

We have no data on treatment of those who reported asthma attacks, and cannot exclude possible confounding by treatment of asthma. Our question phrasing, "Did you ever have attacks of shortness of breath with wheezing (asthma attacks)?" does not automatically imply that asthma was diagnosed by a doctor. Thus, most likely we studied mild disease, and not everyone who reported asthma attacks was also treated for it. In addition, this study was performed in the 1960s and because the inhaled steroids were introduced in the 1970s, the mainstay of treatment will have been -agonists. -Agonists are unlikely to change underlying inflammation and, hence, will not influence the association between eosinophilia with asthma attacks and increased mortality from COPD.

We considered whether several sources of misclassification of the death certificates' diagnosis of mortality from COPD, i.e., differential misclassification, classification of COPD deaths as asthma, or classification of asthma deaths as COPD, could explain these results. Differential misclassification, i.e., specification of the death cause being dependent on the answer to the question about a history of asthma attacks, seems unlikely. Subjects and investigators were both unaware of the future death cause, whereas doctors were unaware of the answer in this epidemiologic investigation years earlier. During our study period, changes in the ICD may have affected death diagnosis. The ICD-9 changes clearly indicate that there has been a movement away from COPD toward asthma (28), thus increasing diagnostic preference for asthma as the cause of death in older people. If this is the case in The Netherlands, former COPD deaths would now be classified as asthma deaths and thus would reduce the association between asthma and COPD mortality toward no effect. Besides, only five subjects died of asthma, three of whom died after the beginning of 1979 when ICD-9 was implemented in The Netherlands. Therefore, this is an unlikely explanation of our results.

Since older asthmatics often have concomitant obstructive airway disease, lower-than-predicted lung function, no allergy, and a smoking history (29), the question arises whether mislabeling of asthma as COPD in the elderly by doctors may have affected our results. Spirometry may not sufficiently discriminate asthma from other forms of airway disease in the elderly (29). From the literature it is known that asthma in the elderly is frequently misdiagnosed or unrecognized (29, 30). One study (31) showed that nearly 75% of newly diagnosed asthmatics over the age of 40 yr previously had a diagnosis of chronic bronchitis or emphysema. Thus, when a doctor is presented with an older patient who smokes and has irreversible airway obstruction, it is likely that the doctor will diagnose COPD, especially when the patient is male (29). Except for these problems concerning the diagnosis of asthma in the elderly, there is a lack of incentive for doctors to record the exact cause of death on the death certificate which weakens the system for reliable cause-specific analyses. Therefore, we cannot rule out the possibility that our data underestimate asthma as the cause of death in favor of COPD.

In our study of all respiratory symptoms, only dyspnea  III and not phlegm, wheeze, or cough was associated with increased COPD mortality. This finding agrees with available literature (32) reporting that phlegm was not associated with increased COPD mortality, and that breathlessness continued to predict COPD mortality after adjustment for lung function (34). An increased risk of COPD in men can be explained by residual confounding of cigarette smoking, and because of physician bias resulting in labeling the diagnosis of COPD more often in men than in women. Smoking habits were associated with increased COPD mortality, in agreement with other studies (10). As previously reported (8, 32), low lung function is associated with increased mortality from COPD. In our study FEV₁ < 80% of predicted showed relative risks up to 9.28 and 7.74 compared with FEV₁ > 100% of predicted for COPD as the primary and combined causes of death respectively. FEV₁ 80 to 100% of predicted was significantly associated with increased mortality from combined causes of COPD. The 137 subjects with COPD as a secondary cause of death died primarily because of ischemic heart disease (45%), other cardiovascular disease (17%), and lung cancer (15%), all known to be associated with smoking (10). Subjects with COPD as a secondary cause of death died at an earlier mean age than subjects with COPD as the primary cause, respectively, 71.8 and 74.1 yr, suggesting the latter survived other death causes. Finally, we found that underweight was also strongly associated with increased COPD mortality, which has been reported by Chailleux and coworkers (21).

In conclusion, positive skin tests are not associated, whereas eosinophilia is associated, with increased COPD mortality, yet only in combination with a history of asthma attacks. Further research should investigate whether this finding is a result of mislabeling of the death cause and subsequently an underestimation of asthma deaths in favor of COPD, or whether it is a real phenomenon. If it is a real phenomenon, our results show that a history of asthma attacks with eosinophilia is a risk factor for COPD mortality in addition to traditional risk factors such as male gender, older age, smoking, low lung function, dyspnea, and underweight which were all associated with increased COPD mortality. This observation is important because of the increase of asthma prevalence in Western countries (35), which in combination with the decrease in cardiovascular mortality, can lead to a further increase in COPD mortality.