INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Wheezing illness is common in adults, with more than 40% reporting some form of wheezing by age 40 (1, 2). Although adult onset disease is likely to account for over half of adult wheeze (3, 4), little is known about its development. Among children, the onset of wheeze is related to factors such as atopy, gender, familial aggregation, and smoke exposure (5). It is not clear whether these factors also influence susceptibility to wheeze in adulthood. Adult onset wheeze has generally been considered to be nonatopic in nature, more common in women, and influenced by smoking habit (1, 6). Some recent evidence suggests that virtually all wheezing illness has an allergic basis, regardless of the age of onset of symptoms (7). Relatively few longitudinal studies have examined the etiology of wheeze which begins in adulthood (1, 2, 9) perhaps because of the heterogeneous nature of adult wheeze, difficulties in distinguishing asthma from chronic obstructive pulmonary disease (COPD), and overlap of these diagnoses (6). In this nested case control study, we examined risk factors for adult onset wheeze in a community cohort of middle aged subjects who had no childhood respiratory symptoms when they were identified in 1964.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

In 1964, the British Medical Research Council Medical Sociology Research Unit conducted a random community sociological survey of the parents of one in five children who were at primary school in the city of Aberdeen, Scotland. Among the 2,511 children aged 10 to 14 whose parents were interviewed, 288 were reported as having wheeze and 2,223 as having no respiratory symptoms. In a further study of those who wheezed, 121 were classified as having asthma and 167 as having wheeze in the presence of upper respiratory infections (13, 14). The outcome in middle age (i.e., age 34 to 40 yr) for the 288 symptomatic subjects together with 167 comparison subjects selected from those who were asymptomatic as children has been reported (15). In 1995, we attempted to contact the remaining 2,056 individuals, now aged 39 to 45 yr, who had no childhood wheeze; 1,799 subjects were traced and screening questionnaires were posted to the 1,758 surviving subjects. There were 1,542 respondents to this follow-up survey, of whom 177 (11.5%) reported adult onset wheeze, defined as wheeze beginning on or after age 15 (4). Two thirds of the respondents were still living in or near Aberdeen. For this nested case control study, all 117 cases with adult onset wheeze who resided locally, and 277 controls randomly selected from among the respondents with no wheeze who lived locally were invited to take part in further investigations. Ethical approval for the study was obtained from the Grampian Health Board and the University of Aberdeen Joint Ethical Committee and all participants gave informed written consent.

Protocol

Cases and controls were contacted in random order over a 7-mo period from April to October 1995. An interview included questions about respiratory symptoms, socioeconomic factors, smoking habit, and family history of atopic disease. Spirometric tests, FEV₁ and FVC, were measured using a portable spirometer (Vitalograph Compact^®; Vitalograph, Buckingham, UK). Each subject was asked to complete a peak flow diary for one week using a mini-Wright^® peak flow meter (Clement Clarke, London, UK). The best of three peak flow readings was recorded twice daily, on rising and at bedtime, before bronchodilator use.

Challenge tests were carried out using methacholine administered by the method of Yan and coworkers (18). Subjects were asked not to use bronchodilator inhalers for a minimum of 8 h before attending and the test was rescheduled for those who reported that they had had a respiratory tract infection within the previous two weeks. Subjects were excluded from the test if their baseline FEV₁ was less than 70% of predicted. Methacholine was administered using DeVilbiss No. 40 hand-held nebulizers (DeVilbiss, Somerset, PA) starting with a dose of 0.016 µmols followed by doubling cumulative doses. Inhalations were continued until a 20% fall in FEV₁ (PD₂₀) occurred or the maximum cumulative dose of 16.38 µmol had been reached.

Skin prick tests were performed using house dust mite (D. pteronyssinus), cat hair, and mixed grass pollens (Dome/Hollister-Stier, Spokane, WA) (19). The negative control was 50% glycerin. A positive test was defined as a wheal diameter of 3 mm or more greater than the negative control measured 10 minutes after inoculation (17). Venous blood samples of 10 ml were taken for allergen specific and total IgE. Specific IgE tests for house dust mite, cat, and mixed grass pollen were performed by a standard radioallergosorbent test (RAST) technique using a paper disk containing relevant allergens (Phadiatop; Pharmacia Diagnostics, Milton Keynes, UK). RAST results for individual allergens were reported as positive when the RAST class was one or greater (i.e.,  0.35 IU/ml) and negative when the RAST response was zero (i.e., < 0.35 IU/ml) (17). Total IgE was determined by paper radioimmunosorbent test (Pharmacia Diagnostics).

Statistical Analysis

FEV₁ and FVC were expressed as percent predicted values on the basis of predicted values for adult lung function (20). Peak flow diary results were expressed as peak flow variability (PFV), calculated as amplitude percentage mean, and were logarithmically transformed to normalize their distribution (21). To permit analysis in the log scale, a constant of one was added to each value of PFV to eliminate zero values. Methacholine responsiveness was expressed as positive for subjects whose FEV₁ fell by 20% from baseline over the course of the test; all others were expressed as negative. Skin and specific IgE tests were expressed as positive if at least one antigen showed a positive result. Total IgE was considered both as a categorical variable, where results greater than 120 IU/ml were considered positive (17), and as a continuous variable, where values were logarithmically transformed to normalize their distribution. A variable summarizing the positive responses on all three measures of atopy (skin, specific, and total IgE) was calculated as the sum of measures positive. For family history of atopic disease, each occurrence of asthma, eczema or hayfever in parents and siblings was counted and a variable categorized as no affected, one affected, and more than one affected relative. Social class, based on subjects own occupation, was defined as manual or nonmanual using a standard classification (22).

The association of wheeze with measured variables was assessed by chi-square, t tests, and analysis of variance as appropriate using SPSS^® 6.0 (SPSS Inc., Chicago, IL) and by multiple logistic regression using STATA^® Release 4 (Stata Corporation, College Station, TX). The independent effects of potential risk factors were examined among all cases with adult onset wheeze (n = 102). In an attempt to address the heterogeneous nature of adult onset wheeze, the risk factors were also examined among three distinct subgroups of cases: those who reported doctor diagnosed asthma (n = 24), those who described cough and phlegm for as much as three months per year in addition to wheeze (n = 31), and those remaining who reported wheeze (n = 47). The chi-square test for trend was applied where appropriate. The goodness of fit of the logistic regression models was assessed using the test described by Hosmer and Lemeshow (23). An adequate fit was obtained for all models reported in this paper. The several measures of socioeconomic status, atopy and family history were highly correlated. Thus, one measure for each of these risk factors (i.e., social class, the sum of positive measures of atopy and the sum of affected relatives) was used to calculate the adjusted odds ratios. Social class was chosen as the primary measure of socioeconomic status because it is the measure most widely used in the United Kingdom.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Of the 394 eligible subjects who were identified for the case-control study, 14 individuals (4 cases and 10 controls) had moved out of the Aberdeen area. Interview data were obtained from 319 (83.9%) of the remaining 380 subjects who we invited to participate (Table 1). The mean age (SD) of those interviewed was 41.8 (1.6) yr for the cases and 41.6 (1.5) yr for the controls. The gender distribution was similar among cases and controls and reflected the composition of the original group: the ratio of females to males was 1.1:1 for the original 2,056 asymptomatic subjects, 1.2:1 for the cases and 1.3:1 for the controls. Subjects who participated did not differ from those who declined to participate in terms of gender, smoking habit or social class. Although fewer subjects performed the tests requested following the interview, spirometry measurements, peak flow diaries, skin tests, and blood tests were completed for approximately 80% of the cases and 75% of the controls. Methacholine challenge data were obtained from over 70% of subjects; eight subjects (five cases and three controls) were excluded from the test because their resting FEV₁ was less than 70% of predicted and one pregnant subject was also excluded.

Wheeze in the previous year was reported by 71% (72/102) of all cases, including 67% (16/24) of those with doctor diagnosed asthma, 84% (26/31) of those with cough and phlegm, and 64% (30/47) of those with other wheeze. The highest proportion of cases who reported activity limitation due to wheeze was observed in the subgroup with cough and phlegm; 35% (11/31) of this group reported activity limitation compared to 17% (4/24) with doctor diagnosed asthma and 15% (7/47) with other wheeze. Only 27% (28/102) of cases were receiving current treatment for wheeze; these included 71% (17/24) of the group with doctor diagnosed asthma and 35% (11/31) of the group with cough and phlegm. None of the cases with other wheeze were receiving treatment.

Results of the ventilatory function tests are shown in Table 2. Cases with adult onset wheeze had significantly lower FEV₁ (% of predicted) values, lower FVC (% of predicted) values, and lower FEV₁/FVC (%) ratios than control subjects; values for both cases and controls were within the predicted range. Among the subgroups, cases with cough and phlegm had the lowest levels of lung function, measurements significantly lower than the controls. Cases with doctor diagnosed asthma and other wheeze had very similar lung function; their measurements were not significantly lower than controls with the exception of FEV₁ (% of predicted) of cases with other wheeze. On the whole, cases showed significantly greater peak expiratory flow variability and reactivity to inhaled methacholine compared to controls. Cases with chronic cough and phlegm had significantly higher levels of peak flow variability than controls and all three subgroups had significantly greater reactivity to methacholine compared to controls.

The associations between adult onset wheeze and the potential risk factors are summarized in Table 3. There was no significant association with gender. The distributions of socioeconomic characteristics according to educational achievement, housing tenure and social class differed significantly between cases and controls. All three measures of socioeconomic status were associated with adult onset wheeze after adjustment for the effects of the other risk factors. There was a significant difference in the employment status of the women: 38 (68%) cases were in full or part time employment compared with 108 (87%) controls (² = 9.32, p = 0.002). Cases were significantly more likely to report respiratory (15 [15%] cases versus no control subjects) or other health problems (31 [30%] cases versus 35 [16%] control subjects) associated with employment (² = 46.54, p < 0.001, df = 2). Nine cases had given up jobs due to respiratory problems at work. Among the 15 cases who reported that they had wheezing problems related to their employment, all but one were in manual jobs, the most common being the fishing industry. These cases reported that cold, damp working conditions and fumes and dust in their work environments tended to aggravate their wheezing symptoms.

Smoking habits differed significantly among cases and control subjects, a greater proportion of cases being current smokers. The number of years smoked was similar among cases (22.2 yr, SD 9.1) and control subjects (20.4 yr, SD 8.0) (t = 1.39, p = 0.17); former smokers had smoked for a mean duration of 14.4 yr (SD 8.4) and current smokers for 25.5 yr (SD 4.6). Calculation of pack-years (cigarettes per day × duration in years/20), which was only possible for current smokers, showed that cases had a significantly greater cumulative cigarette exposure; 31.8 pack-years (SD 17.3) for the cases compared with 23.5 pack-years (SD 11.6) for the controls (t = 2.81, p = 0.006). Among the 24 subjects with adult onset wheeze who were former smokers, there was a strong correlation between age at onset of wheeze and age when subjects quit smoking (r = 0.65, p = 0.001). After adjustment for the effects of the other risk factors, current smoking was independently associated with adult onset wheeze and there was a significant increasing trend in the odds ratios with increase in duration of smoking (² trend = 6.33, p = 0.012).

Cases with adult onset wheeze displayed significantly higher levels of specific and total IgE, but allergy skin test reactivity was similar among the groups. The most common response observed in the tests was to house dust mite followed by grass and then cat, for both cases and controls. In both univariate and multivariate analyses, there was no relationship between skin test positivity and wheeze. Positive responses for specific and total IgE measurements were associated with adult onset wheeze independent of the other risk factors; when these measures were included in the model simultaneously, only the effect of total IgE remained significant. There was considerable overlap in the specific and total IgE results: 33% of subjects who had positive specific IgE results had positive total IgE results; and 67% of subjects who had positive total IgE results had positive specific IgE results. Analyses using atopy categorized as the sum of positive measures showed that a positive response on all three measures was independently associated with wheeze, but a positive response on any one or two measures did not achieve significance.

A significantly higher proportion of cases had parents or siblings with a history of asthma or hayfever than controls. The proportion of subjects with a family history of eczema was not significantly different between the groups. The adjusted odds ratios showed associations with a family history of asthma or hayfever independent of the other risk factors. A stronger independent effect was observed when more than one relative (i.e., parent or siblings) had a history of atopic disease and there was a significant trend with increase in number of affected relatives. The associations between family history and adult onset wheeze were independent of family size.

Table 4 shows the relationships between the three distinct subgroups of cases and the potential risk factors, mutually adjusted for all factors. The mean age (SD) and ratio of females to males was 42.3 (1.6) yr and 1.7:1 for doctor diagnosed asthma, 42.1 (1.5) yr and 1.1:1 for cough and phlegm, and 41.4 (1.6) yr and 1.1:1 for other wheeze. There was no association with gender for any of the subgroups. Social class was associated with the subgroups which had not acquired a diagnosis of asthma, that is cough and phlegm and other wheeze. There was a strong relationship between smoking habit and cough and phlegm. Atopy categorized as the sum of positive measures was associated with a diagnosis of asthma but not with other wheezing subgroups. However, log total IgE, included in the analyses in place of the summary variable, was independently associated with doctor diagnosed asthma (OR = 1.93, 95% CI = 1.29 to 2.89) and chronic cough and phlegm (OR = 1.67, 95% CI = 1.17 to 2.36); the association with other wheeze (OR = 1.31, 95% CI = 0.99 to 1.72) did not achieve significance. Family history of atopic disease was related to all three subgroups; significant trends were observed in the odds ratios with increased number of affected relatives.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We defined cases based on self-reported wheeze rather than specific clinical or diagnostic labels. Our case group is therefore likely to be heterogeneous, including subjects with asthma, COPD and possibly other obstructive airways disorders. We believe, however, that our broad case definition is informative and valid and that use of a more restricted definition such as doctor diagnosed asthma or COPD would be associated with several limitations. These include concurrent diagnoses of more than one obstructive airways disease (6), diagnostic bias in relation to perceived risk factors (1, 24), underdiagnosis of characteristic symptoms (1, 3), and underreporting of diagnoses by subjects (10). We have examined risk factors for wheeze as reported by subjects screened in a population study; the inherent heterogeneity of our case definition was explored by examining the relative importance of risk factors in subgroups of cases.

The objective measures of ventilatory function observed in this study support the validity of our case control definitions; compared with control subjects, cases with adult onset wheeze had significantly greater airflow obstruction, peak flow variability and reactivity to methacholine. As might be expected, cases with cough and phlegm had the lowest levels of lung function among the subgroups. These cases were the only subgroup to have significantly higher peak flow variability than controls suggesting that the relatively high levels of activity limitation due to wheeze reported in this subgroup may reflect underrecognition of asthma-like symptoms and undertreatment. Although the cases with a diagnosis of asthma had similar lung function and reactivity compared with the remaining wheezing cases, this may reflect the effect of treatment in the asthma subgroup. It would appear that the subgroup with other wheeze may not have had symptoms or lung function impairment that was severe enough or characteristic enough of asthma to lead to clinical diagnosis or treatment.

Risk factors for adult onset wheeze identified in our study included social class, smoking habit, atopic status and family history of atopic disease. Our finding that the relative importance of the risk factors varied in the subgroups is compatible with reports of others (6, 9, 10, 25). Several prospective longitudinal studies in the United States and one in the United Kingdom have examined the development of adult onset wheeze in previously asymptomatic persons. These studies have incorporated various case definitions based on symptoms or diagnostic labels; their findings are contrasted with ours in Table 5.

Gender was not associated with adult onset wheeze among all cases or the subgroups in this study. Current epidemiological evidence regarding the effect of gender on development of wheeze in adulthood is conflicting. While this may reflect the use of different case definitions or diagnostic bias, it is possible that the overall effect of gender is relatively weak and not always independent of other risk factors. The association between asthma diagnosed in adulthood and female gender observed in the Tucson population studies was not adjusted for the effects of other risk factors (1, 6). McWhorter and coworkers reported that being female was an independent risk factor for adult onset doctor diagnosed asthma but not COPD (10). The findings of both these studies may reflect the tendency of physicians to diagnose respiratory complaints as asthma in females and as COPD in males. Although Strachan and colleagues observed a weak independent association between adult onset wheezing illness (including asthma and wheezing symptoms) and female gender, their finding should be interpreted with some caution because it is based on data from 31% of their original cohort (2).

Socioeconomic factors, as measured by educational achievement, housing tenure and social class, were independently associated with adult onset wheeze. In the subgroup analysis, social class was related to cough and phlegm and other wheeze, but the association was attenuated with doctor diagnosed asthma. The reason for this increased risk in lower socioeconomic groups, even after adjustment for smoking, atopy and family history, is uncertain; environmental exposures common to the manual class, occupational factors, dietary habits, or residual smoking effect may have an influence. An association between low social class and adult wheeze, independent of smoking, has been reported previously (26). In another study, the relationship between social class (as measured by father's social class) and adult onset asthma or wheeze disappeared after adjustment for smoking (2). There is some evidence that adults in lower socioeconomic groups are less likely to receive a diagnostic label of asthma (24). This may be one explanation for the attenuated effect of manual class in our subgroup of cases with doctor diagnosed asthma. However, low income, a variable similar to social class, was an independent risk factor for both adult onset diagnosed asthma and COPD in a prior study (10).

Current smoking was independently associated with adult onset wheeze. Among the subgroups, smoking habit was strongly associated with chronic cough and phlegm, whereas it was not associated with doctor diagnosed asthma or other wheeze. The importance of smoking as a risk factor in adult populations appears to be influenced by the definition of wheezing illness used. Several studies reported that smoking was a risk factor for the development of wheeze in adulthood (2, 11, 27). However, studies which attempted to distinguish asthma from other forms of wheeze, reported that asthma was not associated with smoking habit, in sharp contrast to COPD (10) and chronic bronchitis (12). This is compatible with our subgroup analysis in which smoking habit was strongly associated with cough and phlegm, but not with doctor diagnosed asthma. The apparent lack of relationship between smoking and diagnosed asthma may be influenced by a tendency of physicians to diagnose smokers with wheeze as having chronic bronchitis or emphysema rather than asthma. In addition to diagnostic bias, selection bias may operate in determining that individuals with sensitive airways are less likely to become regular or heavy smokers and that smokers who develop symptoms tend to quit (12). In our study, there was a correlation between onset of wheeze and cessation of smoking, suggesting that smokers may quit when they develop symptoms because they believe that smoking caused or exacerbated their symptoms (12).

Our finding that atopy was independently associated with adult onset wheeze does not support the traditional division of wheezing illness into extrinsic (allergic) disease as the predominant form in children and intrinsic (nonallergic) disease as the predominant form in adults. Rather, our data are consistent with more recent evidence suggesting that an underlying atopic state is present in most wheezing illness. Our subgroup analysis showed that atopy, categorized as the sum of measures positive or total IgE, was associated with the acquisition of a diagnosis of asthma in adulthood. Total IgE was also associated with chronic cough and phlegm. There was no significant relationship between atopy and other wheeze. Using case definitions based on symptoms, lung function and diagnoses, several studies have demonstrated that the relationship between atopy and wheezing illness is independent of gender, age and smoking habit, factors known to influence the expression of atopy. In the Tucson population, the onset of asthma before age 40 was associated with both skin test reactivity and eosinophilia, while onset after age 40 was associated with eosinophilia and not skin test reactivity (1). Although this was originally cited as evidence for extrinsic and intrinsic divisions, subsequent reports from the same population showed that asthma was closely related to total IgE levels within all age groups even after gender differences, the degree of skin test reactivity, and smoking history were accounted for (7, 28). In a cohort of men in Boston, the onset of wheeze and chronic cough and phlegm after age 21 was associated with total IgE, but not with skin test reactivity, after adjustment for age and smoking status (9). In other national U.S. and British cohorts, adult onset wheeze has been shown to be related to self-reported markers of atopy; i.e., a history of allergies, hayfever, and eczema (2, 10).

In this study, a positive response to all three measures of atopy (rather than only one or two) was associated with adult onset wheezing. Although this suggests that all the measures contributed to the risk of wheezing illness, these markers may not be interchangeable indicators of atopic status. In all analyses, both univariate and multivariate, skin test reactivity showed no relation to wheeze in all cases or in the subgroups. Although specific IgE positivity was independently related to wheeze in all cases, this relationship was attenuated in analysis of the subgroups. In addition, after the level of total IgE was accounted for, specific IgE positivity no longer showed an independent relation to wheeze in all cases. Total IgE appeared to be the measure of atopy which was most closely related to adult onset wheeze; significant associations were also noted for the subgroups with doctor diagnosed asthma and chronic cough and phlegm. Risk of wheeze among all cases increased significantly with increasing logarithmic levels of total IgE above 3 IU/ml (² trend = 14.7, p < 0.001) (7, 8). It is possible that the allergens commonly used in skin or specific IgE tests may not adequately represent the allergic stimulation responsible for the association observed between total IgE and adult onset wheeze in this and other studies (7, 9). Potential allergic stimuli in adult onset disease may include bacteria in the respiratory tract or occupational exposures. Another possibility is that different categories of respiratory symptoms (e.g., asthma and hayfever) are associated with different immunologic factors which are reflected by distinct patterns of atopic expression (7, 9). The high correlation between the specific and total IgE results may make it difficult to look at the separate effect of the two measures; collinearity may be responsible for the attenuation of the effect of specific IgE observed in this study.

A family history of asthma, eczema or hayfever was the strongest independent risk factor for adult onset wheeze. A family history of all three atopic diseases was more important risk factor than a family history of asthma alone suggesting that the concomitant inheritance of a predisposition to all three atopic diseases enhances the likelihood that symptoms will be expressed. By virtue of the fact that our subjects knew they were participating in a respiratory study, they may have been more likely to recall a history of asthma in their parents or siblings; this should not have influenced their recollection of eczema or hayfever. The influence of study participation would have affected cases and controls in a similar manner making it unlikely that a systematic bias produced gross overreporting in this study.

Although it is well known that wheezing in children is often associated with a family history of atopic disease, few studies have examined the relationship between family history and adult onset wheeze. The development of wheeze in adulthood was related to maternal hayfever (as reported by the parent) in one study (29) and to parental asthma (as reported by the subject) in another (30). A tendency to develop COPD and airflow obstruction appears to aggregate within families (31); perhaps this association is linked to a shared allergic predisposition. Our observation that family history of atopic disease was associated not only with doctor diagnosed asthma, but also with chronic cough and phlegm and other wheeze, suggests that different forms of adult onset wheeze may have a common allergic basis. The expression of symptoms as asthma or COPD may depend on an interaction between the atopic diathesis and environmental exposures. Our findings are consistent with the hypothesis that atopy is one of the major determinants of various forms of obstructive airways disease (32, 33).