INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Repeated peak expiratory flow (PEF) measurements are widely used in clinical and epidemiological settings for assessing the severity of asthma and as an indicator of bronchial hyperreactivity. In addition, they are also used to measure changes that occur in pulmonary function related to treatment, or to environmental exposures (1, 2).

The variability of PEF has two dimensions: diurnal (within-day) and day-to-day variation. These two represent different aspects of the disease, acute and chronic, and may therefore have different predictors. On the other hand, the same predictors (e.g., skin prick test results) may have different relative importances for the two measures of PEF variability.

Children with asthma have a greater variation in PEF than children with no respiratory symptoms (3). Among clinically diagnosed asthmatic children, an increased variation in PEF has been reported to be associated with symptom scores, daily medication use, lowered lung function, and bronchial hyperresponsiveness (2, 6). Chronic cough is sometimes considered to be an under-diagnosed form of asthma (9). Few studies, however, have compared variation in PEF between children with asthmatic symptoms and those with cough as their only chronic respiratory symptom. Moreover, few studies have investigated the relationship between PEF variability and, respectively, chronic respiratory symptoms, lung function, and atopy in population samples of children with chronic respiratory symptoms.

In epidemiological studies, questionnaires are often used to characterize subjects or subgroups of potentially sensitive subjects, for example, based on respiratory symptoms. It is important to assess how well that characterization relates to peak flow variability as an objective marker or bronchial lability. Therefore, we investigated how chronic respiratory symptoms reported in a questionnaire, skin prick test results and spirometric lung function predicted diurnal and day-to-day variability in PEF during a 2- to 3-mo daily follow-up. Data were collected in the framework of the PEACE study (Pollution Effects on Asthmatic Children in Europe) (10), which was conducted in 14 different areas in Europe, including five in central and eastern Europe. Previously, we showed that in the Finnish panels, asthmatic symptoms and allergy to house dust mite were related to reduced MMEF (maximal mid-expiratory flow) values (11). This study extends these observations to lung function variability, using data from the PEACE study as a whole.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Locations

The study was carried out within the framework of the PEACE study on effects of air pollution on respiratory health of children. Fourteen study centers in 10 countries took part in this European collaboration. The participating centers by the location of the urban study area were Amsterdam (The Netherlands), Kuopio (Finland), Oslo (Norway), Berlin and Hettstedt (Germany), Pisa (Italy), Athens (Greece), Cracow and Katowice (Poland), Prague and Teplice (Czech Republic), Budapest (Hungary), Umeå and Malmö (Sweden). In each center, the same study protocol was followed (10). In Hettstedt, the procedure used for selection of subjects was somewhat different, however. Therefore, in the present analyses data from 13 study centers are used, excluding Hettstedt.

Study Design

In each center, there were two panels of 6- to 12-yr-old children with chronic respiratory symptoms: an urban and a control panel. The children were selected by a parent administered screening questionnaire. The screening questionnaire was administered to representative population samples of school-age children living in the areas that were selected for the air pollution study. Generally, study areas were selected on the basis of vicinity to the participating study center, and expected air pollution contrasts between urban and control site. Children in the urban panel were living in an urban area and children in the control panel were living in a suburban or rural area. Having urban as well as control areas was based on the aim of the PEACE study to examine effects of air pollution on respiratory health. During the autumn of 1993 and the winter of 1994, the children were characterized with skin prick tests and spirometric lung function measurements. During the winter of 1993-1994, the children measured peak expiratory flow (PEF) every morning and evening and kept with help of their parents a daily diary on respiratory symptoms for two to three months (10).

Subjects

Of approximately 3,000 6-12-yr-old children eligible to enter the study and asked to participate, 2,112 children were characterized with skin prick test and spirometric lung function measurements. A total of 2,152 children started measuring daily PEF and keeping a diary on respiratory symptoms. Of these, 1,854 (86%) children had valid PEF and symptom data on more than 60% of the days and these children were included in the present analyses (Table 1, Table 2). The 1,854 children with valid data were not significantly different from the original sample in any of the descriptive variables shown in Tables 1 and 2.

Screening Questionnaire

The screening questionnaire used in the PEACE study was an adapted version from previously used questionnaires: a World Health Organization questionnaire for assessing respiratory symptoms in children (12) and a questionnaire developed in the University of Groningen, The Netherlands. The latter was based on the American Thoracic Society questionnaire for children (10, 13). In addition, the question on nocturnal cough was taken from a study conducted in New Zealand (10, 14). A child was considered eligible to enter the study if there was a positive response to at least one of the following questions:

1. Has your child been bothered in the past 12 mo by a wheezy chest, apart from colds?
2. Has your child been bothered in the past 12 mo by attack of shortness of breath with wheezing?
3. Has your child had dry cough at night in the past 12 mo, apart from coughing with a cold or chest infection?
4. Has a doctor ever said your child had asthma? (10).

Skin Prick Tests

In all centers, skin prick tests were done using the ALK skin prick test system (ALK Laboratories, Denmark). The allergens tested in all centers were birch (Betula verrucosa) and timothy grass (Pheleum pratense) pollen, cat fur (Felis catus), and house dust mite (Dermatophagoides pteronyssinus) (10). Histamine hydrochloride (10 mg/ml) and glycerol (50%) were used as positive and negative controls, respectively. In addition, two local allergens were tested in each center. They were dog epithelial dander and cladosporium in Amsterdam, Oslo, Umeå and Malmö, dog epithelial dander and mugwort pollen in Kuopio, alternaria and plantago in Berlin, alternaria and ragweed in Budapest, alternaria and parietaria in Pisa, olive and parietaria in Athens, grass mix pollens and dog epithelial dander in Cracow and Katowice, and grass mix pollens and mould mix in Prague and Teplice. The allergens were ordered and distributed by the coordinating center. Reactions were read 15 min after allergen application by outlining the circumferences of all resulting welts on the skin, and by transposing them to a data collection sheet with transparent tape. A mean weal diameter of more than 2 mm was regarded as a positive result (10). If there was no reaction to the positive control or a reaction more than 1 mm to the negative control, the results were excluded from the analyses.

Lung Function Tests

The spirometric lung function measurements were done according to the recommendations of the European Community for Coal and Steel (10, 15). The subjects were seated and used a noseclip. The largest values of forced vital capacity (FVC) and forced expiratory volume in one second (FEV₁) were selected from a minimum of three valid expiratory recordings. The largest maximal mid-expiratory flow (MMEF) was selected from a recording with the FVC value within 5% of the largest FVC. All spirometric results were corrected to body temperature, atmospheric pressure, and saturation with water vapor (BTPS). Predicted values based on sex and height of the subjects were calculated according to Zapletal (16).

The spirometer used in the different centers were not identical, but fulfilled the technical requirements of the European Community for Coal and Steel (15).

PEF Measurements

The children were followed-up for 2-3 mo during winter 1993-1994. They measured peak expiratory flow rate (PEF) every morning and evening three times in standing position with a mini-Wright Peak Flow meter (Airmed; Clement Clarke International Ltd, Essex, UK) before taking any respiratory medication. All three PEF readings were noted in a diary, and the largest of these three readings was used for the analyses. The children also kept a daily diary on respiratory symptoms with help of their parents (10).

Definitions

Children who had ever had asthma diagnosed by a doctor, and those who had suffered from wheezing or attacks of shortness of breath with wheezing during the previous 12 mo were defined as having asthmatic symptoms. Children who did not have these asthmatic symptoms but had suffered from dry cough at night not associated with colds were defined as having cough alone. Atopy was defined as having at least one positive reaction in the skin prick tests.

Stastical Analyses

Data were analyzed by using the statistical package SAS /STAT^® (SAS Institute Inc., Cary, NC) (17). The variation in PEF was assessed in two ways in order to examine both diurnal and day-to-day variation in PEF. To describe the diurnal variation, a mean daily variation in PEF (%) was calculated for each child. First, daily variation was calculated as (morning PEF  evening PEF)/[(morning PEF + evening PEF)/2] × 100, and these daily values were then averaged over the study period (2). To describe day-to-day variation, a coefficient of variation (CV) of morning PEF (%) was calculated as (SD of morning PEF values)/(mean of morning PEF values) × 100 for each child (2, 18). Instead of evening PEF, morning PEF was chosen, because it was thought that morning PEF values were less affected by bronchodilating medication possibly used during the day. These two parameters were chosen as they have been reported to provide a reasonable approach to expressing PEF variability for epidemiological purposes (18). To account for possible learning effects, at least the two first days of PEF measurements of each child were excluded from the analyses (3).

Associations between skin prick test results, spirometric lung function, chronic respiratory symptoms and variation in PEF were analyzed using analysis of covariance (proc GLM). The urban and control panels of each center were combined to increase the power of the study, as preliminary data analyses did not suggest that there were significant differences between panels studied by the same center. Two models were constructed. In the first model we adjusted for sex, age, and study center, and each personal characteristic under study was added to the model one at a time. In this model, the estimated effects of atopy, chronic respiratory symptoms and lowered lung function on variation in PEF may be overestimated, as these characteristics may relate to one another. The second model was a full model including not only age, sex, and study center, but also atopic status (defined as being atopic or not), symptom status (defined as having asthmatic symptoms or not), and the level of lung function, expressed as MMEF as % of predicted. Effects of these variables on PEF variability were expressed as absolute differences in % variability: an absolute difference of 2% in morning PEF CV means an increase in the CV from 8% to 10%, which is equivalent to a relative increase of 25%.

As residuals were found not to be normally distributed in the models used, the analyses were repeated with log transformed variables of variability in PEF. In these models, residuals were found to be normally distributed, but the results remained essentially unchanged. Therefore, only nontransformed results are shown.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The mean daily variation in PEF varied between the study centers from 4.4% to 10%. The mean CV of morning PEF varied from 5.5% to 10.8% (Table 2). The correlation coefficient between the mean daily variation in PEF and the CV of morning PEF was 0.76 (p < 0.0001). There were no significant differences between boys and girls in the mean daily variation in PEF or in the mean CV of morning PEF. Both the mean daily variation in PEF and the CV of morning PEF decreased with age.

Children with asthmatic symptoms had a greater variation in PEF than children with cough alone (Table 3). The absolute difference, about 1%, was greater in mean CV of morning PEF than in mean daily variation in PEF. Children with doctor diagnosed asthma had a greater variation in PEF than children without doctor diagnosed asthma. This was also found when we restricted the analysis to children with asthmatic symptoms. In addition, the more symptoms were reported in the screening questionnaire the greater the variation in PEF was. The children for whom all four symptoms asked in the questionnaire were reported, had about a 2% greater CV of morning PEF than those with only one symptom, which is a 20-40% increase over the mean morning PEF CVs, which ranged from about 5 to about 10% (Table 2).

Atopy and especially positive skin test reactions to indoor allergens (house dust mite, cat) were associated with an increased variation in PEF (Table 4). Although there was a significant positive trend in the association between the number of positive skin prick test results and variation in PEF, it was those with 5 or 6 positive skin prick test reactions who clearly had a greater variation in PEF than the others. They had about a 3% greater mean daily variation in PEF and about a 4% greater mean CV of morning PEF than nonatopic children. When stratifying the analyses by symptom status, the association between skin prick test results and the variation in PEF was observed only among children with asthmatic symptoms but not among children with cough alone. Among asthmatic children, the adjusted associations between skin prick test results and the variation in PEF were similar to those in Table 4.

A low level in spirometric lung function (as % of predicted), especially in FEV₁ and MMEF, was associated with a greater variation in PEF (Table 5). When using the full model with lung function as a continuous variable, there was a 0.36% difference in mean daily variation in PEF and a 0.29% difference in mean CV of morning PEF for the interquartile range of FVC. The respective figures were 0.70% and 0.68% for FEV₁ and 0.86% and 1.09% for MMEF. Stratification of the analysis by symptom status did not change the associations between lung function and the variation in PEF. Relationships between PEF variability and, respectively, symptom status, atopy, and lung function level were mostly not different between girls and boys, or between children of different ages. There was some indication that the difference in morning PEF CV between children with asthmatic symptoms, and children with cough only was bigger in girls than in boys (difference was 1.03% in girls (with an SE of 0.36%), 0.52% in boys (SE 0.34%). The effect of atopy on morning PEF CV was somewhat bigger in the younger half of the children (0.96% with an SE of 0.33%) than in the older half of the children (0.42% with an SE of 0.31%).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this large epidemiological study among children with chronic respiratory symptoms, we have documented that Europe-wide, clear differences in PEF variability exist between the three main groups in the study (those with doctor diagnosed asthma, those with symptoms of asthma but no doctor's diagnosis of asthma, and those with nocturnal cough but no [other] symptoms of asthma). The large size of the population also allowed us to document relationships between PEF variability and different aspects of atopy, as well as different measures of lung function level in detail. The associations were observed both in the diurnal and day-to-day variation in PEF. Differences in PEF variability, expressed as absolute differences of up to a few % points were found, amounting to relative differences of up to 50%. On a group level, such differences are clearly important.

In the present study, PEF was monitored over the study period with measurements at home twice a day. While this in entirely dependent on subjects, there may be problems with compliance. It has been reported that errors in reading and transcribing peak flow values occur in a subset of asthmatics (19). In addition, it has been suggested that PEF monitoring using ordinary peak flow meters for assessment of work-relatedness of asthma has limitations and is not reliable (20). As quality control in the present study, however, all subjects included in the analyses has at least 60% of the possible days valid diary data, i.e., both PEF values and symptom recordings were reported for the same day.

A mean diurnal PEF variation of 8% has been quoted as a normal level in children (3). In the present study among symptomatic children, the average level of mean diurnal variation in PEF exceeded 8% only in one study center. This might reflect a good control due to effective therapy of the disease achieved by, e.g., proper medication. In subjects with only cough symptoms, who have closer to normal values of pulmonary function and are unlikely to be medicated, the mean daily variation in PEF (5.6%) was also below 8%. On the other hand, the PEF measurements were done twice a day, in the morning and the evening at bedtime. Peak expiratory flow has a circadian rhythm. It is at lowest in the early morning and at highest in the afternoon or early evening, declining after that again towards the night (1, 3). By having only two measurements a day, the level of diurnal variation in PEF may be underestimated (21).

The level of PEF variability observed in present study is in agreement with previous studies, in which twice a day measurements of PEF were used. Among 7- and 8-yr-old children with wheeze or cough symptoms, followed for a year, a mean diurnal variation in PEF of 4.6% and a CV of morning PEF of 13.9% have been reported (22). Roemer and coworkers (23) reported a CV of morning PEF of 9.8% among 6-12-yr-old children with chronic respiratory symptoms followed for three months. Among 5-12-yr-old children with asthma, a geometric mean of mean diurnal variation in PEF of 7.1% has been reported (2). The respective figure in the present study was 5.9% for the asthmatic children and 5.1% for the children with cough alone.

The number and severity of chronic respiratory symptoms reported in the screening questionnaire predicted the variability of PEF. Of the symptoms asked, ever doctor diagnosed asthma was the best predictor for an increased variation in PEF. The absolute differences were greater in the mean CV of morning PEF than in the mean daily variation in PEF. This may reflect that the CV of morning PEF is a better indicator of the long-term status of the disease than the mean daily variation in PEF and it may describe the same as chronic respiratory symptoms asked in the questionnaire.

Atopy was associated with a greater variability in PEF. Of the allergens tested, a positive reaction to dust mite was the best single predictor for an increased variation in PEF. This is in agreement with previous knowledge that the risk of asthma is primary associated with sensitization to indoor allergens (house dust mite and cat) (24, 25). A positive skin prick test reaction to pollen was not as strong a predictor for an increased PEF variability as such a reaction to indoor allergens. This may indicate that the allergen must be present all the time to be able to affect airways and further the variability in PEF. On the other hand, the follow-up was done outside the pollen season. The association between allergy to pollen and PEF variability might have been different if the follow-up was during the pollen season.

The association between atopy and PEF variability was observed only among children with asthmatic symptoms, but not among children with cough as their only respiratory symptom. This is supported by Clough and coworkers (22) who also observed in their study among children with wheeze or cough symptoms that both atopy and wheeze were associated with a significantly greater diurnal variation in PEF and there was a significant interaction. These observations confirm that children with dry cough as their only chronic respiratory symptom seem to differ clearly from children with asthmatic symptoms, concluded also by Wright and coworkers (24). In the present study, they also had a lower variation in PEF than children with asthmatic symptoms. Previously it has been observed that they are less frequently atopic and their MMEF values are larger compared with asthmatic children (11, 22).

Spirometric lung function, expressed as % of predicted, was negatively associated with variation in PEF: the better spirometric lung function the smaller variation in PEF. The effect was the smallest for FVC. Maximal mid-expiratory flow (MMEF) has been suggested to be a sensitive index of airway obstruction (27). MMEF is effort independent and measures flow predominantly in the peripheral airways, whereas FEV₁ measures airflow predominantly in the central airway. In the present study, the difference in variation in PEF was larger for an intequartile range of MMEF than that of FEV₁. This is in agreement with previous observations, in which MMEF was shown to be lower among asthmatic children compared with children with cough alone, while no difference in FEV₁ was observed (11).

Model specification had little effect on associations between chronic respiratory symptoms, skin prick test results, spirometric lung function, and PEF variability indicating that these are independent predictors of variation in PEF. Compared with the model adjusted only for age, sex, and study center, using the full model affected only the effect estimates of chronic respiratory symptoms. Despite that the associations between chronic respiratory symptoms and variation in PEF remained significant.

In conclusion, chronic respiratory symptoms reported in a screening questionnaire, spirometric lung function, and skin prick test results among asthmatic children predicted variation in PEF measured during a 3-mo follow-up. The difference in morning PEF CV between children with asthmatic symptoms and children with cough only was somewhat bigger in girls than in boys. The effect of atopy on morning PEF CV was somewhat bigger in young than in older children.