SUBJECTS AND METHODS

The study took place at the Children’s Asthma Centre, Leicester Royal Infirmary, UK. Forty six children and their parents were recruited; 14 were from an asthma clinic and 32 from local general practices. The children were aged 7–14 years of age, had a clinical diagnosis of asthma, were taking some form of regular medication, and had had no severe exacerbations or changes to their preventer therapy in the four weeks prior to the test. Twenty four subjects volunteered to return for a second identical set of tests (within eight weeks of the first), which enabled repeatability to be tested.

Both parents and children gave written informed consent. The study was approved by the Leicestershire Research Ethics Committee.

Measurement of perception

A 10 cm visual analogue scale (VAS) was constructed with “very bad” and “no problem at all” marked at the extremities. Children or parents were separately asked “Can you mark on the line how your (your child’s) asthma was last week when you (your child) did exercise?”. In order to determine the perception of asthma severity immediately before and after the exercise challenge test, the question “Can you mark on the line how your (your child’s) asthma is now?” was used. The questions and end points were non-specific, in that they did not enquire about a particular symptom or sensation of EIA, as interpretation might depend on the child’s understanding or knowledge. The authors based the questions on pilot work to investigate the terminology used by children.

Perception was also measured on a five point categorical (Likert) scale, to determine the level of “bother” (again, not symptom specific) associated with EIA. The questions were exactly the same for both children and parents: “How bothered were you (was your child) by asthma after the running test?”—extremely bothered, bothered a lot, bothered a bit, hardly bothered at all, not bothered. This question was only asked after exercise.

The following potential confounders of the association between perception and physiological response to exercise were recorded prior to the exercise challenge: age, gender, asthma severity (by British Thoracic Society (BTS) step²³), medication, habitual exercise, attitude towards exercise, and competitiveness. Attitude towards exercise was derived from a validated questionnaire²⁴ using the question “I think that doing exercise long enough to get sweaty a few times a week, during my free time is: (i) very good fun, fairly good fun, neither good fun nor unpleasant, fairly unpleasant, or very unpleasant; (ii) very unhealthy, fairly unhealthy, neither healthy nor unhealthy, fairly unhealthy, or very unhealthy”.

Level of habitual exercise was assessed by asking parents/children to identify which activities were done over the past week, and their frequency, duration, and intensity. An intense activity was one that raised the heart rate and caused sweating. The majority of children did a large number of activities, so categorisation by frequency and time was not possible. Categorisation by intensity included three categories: no intense activity, some intense activity, and all intense activity. Parents and children who attended the laboratory completed the questionnaire and each of the perception scales separately, without collusion at any stage.

Exercise challenge

The exercise challenge was a six minute run on a treadmill (Quinton Q50, USA) at a speed of 6–8 km/h, and a gradient of 8–10% to achieve a heart rate of 85% of the maximum predicted value [220 − age (y)], by the end of the exercise period. The laboratory was air conditioned (19–21°C; relative humidity <50%) during all exercise tests. Measurements of lung function, including PEF, and several spirometric indices were made from full forced vital capacity (FVC) manoeuvres using an electronic spirometer (Vitalograph Compact Spirometer, Ireland), calibrated before the arrival of each subject using a 2 litre syringe (Jaeger). Lung function was measured before (best of three blows according to ATS criteria²⁵) and at 1, 3, 5, 10, 20, and 30 minutes after completing exercise (single blows). The FVC manoeuvres were carried out with the child sitting and with a nose clip. Neither the child nor parents saw the spirometry results. The level/severity of EIA was expressed as the maximum change after exercise from the baseline value of lung function (%). Children whose FEV₁ fell by 13% or more were classified as having exercise induced asthma.⁵

Following the test, after each set of forced vital capacity manoeuvres children were asked: “Describe in your own words how you are feeling now”, to capture qualitatively their perception of symptoms.

Once the FEV₁ had returned to within 5% of baseline, parents and children completed the VAS and Likert scales for a final time. If a child’s FEV₁ had not returned to within about 5% of baseline by 30 minutes, 400 μg of salbutamol was administered by metered dose inhaler and spacer.

Statistical methods

At least 14 subjects were required, in order to have 80% power to detect a true correlation between perceptual score and physiological measures of EIA of 0.7. However, the confidence intervals are wide (0.3 to 0.85). Thus 40 subjects were selected as the sample number to allow for defective data.

Absolute VAS scores were log transformed to achieve normality, as there was positive skewing towards the “no problem at all” end. The change in VAS after exercise was not normally distributed, and so we used the change in log VAS score after exercise, which was normally distributed.

Separately for children and parents, linear regression analysis was used to compare the change in log VAS scores with the change in lung function after exercise. Absolute log VAS scores after exercise were also compared to change in lung function. The absolute log VAS scores for EIA perceived in the previous week were also compared to the severity of EIA in the laboratory.

Parent and child VAS scores before and after exercise as well as change in VAS on exercise were compared by linear regression. Kappa was used to compare parents’ and children’s categorical scale scores. The effect of confounding factors was to be analysed by multiple regression. To evaluate repeatability in symptom perception, differences in the change in log VAS scores between the first and second visit were regressed against differences in the change in lung function.

To validate the VAS, the absolute VAS score after exercise for both parents and children was compared to the categorical scale by linear regression. Repeatability was assessed using the technique of Bland and Altman. Analysis was done with SPSS version 10.

RESULTS

Forty six children were recruited; results from only 43 were analysed, as two children were not on regular medication and one had extremely variable spirometry results, which made interpretation difficult. Thirty six completed the exercise test satisfactorily on the first visit; three children ran for under six minutes because of severe bronchoconstriction, and four could not achieve target heart rate but were still included for analysis as they appeared to be exerting maximum effort while running. On the second visit all 24 children completed the exercise test satisfactorily. The first and repeat visit groups did not differ in age, baseline FEV₁, or features of asthma (table 1). Overall, 60% developed EIA on test 1 and 63% on test 2. The mean fall in FEV₁ after exercise was 22% and 21% respectively.

Visual analogue scale

Using the VAS, children’s perception was related weakly to change in FEV₁ after exercise (r² = 0.1, p = 0.04; fig 1), but not to change in PEF, FEF₅₀, or FEF_25–75. Reported severity of EIA (by VAS) over the previous week was unrelated to any measure of EIA.

Parents’ perception, measured either by change in log VAS scores (r² = 0.03; p = 0.28) or by absolute log VAS (r² = 0.06; p = 0.09) after exercise, was unrelated to any measure of EIA, as was parents’ reports of their child’s EIA over the previous week.

Parents’ VAS related poorly to child’s VAS before (r² = 0.002; p = 0.78) and after exercise (r² = 0.06; p = 0.12). Furthermore, change in parent VAS score related poorly to change in child VAS score after exercise (r² = 0.03; p = 0.28).

Likert (bother) scales

There was very poor agreement between parents’ and children’s categorical “bother” scores (κ = 0.095).

However, the “bother” and VAS sores were highly correlated for children (r² = 0.50; p < 0.001; fig 2), and for parents (r² = 0.58, p < 0.0001). There was a poor relation between the “bother” scale and EIA or lung function. For children who had marked “not bothered” and parents who marked “bothered a bit” the mean fall in FEV₁ was 46%.

Qualitative descriptors

Children’s free descriptors recorded at five minutes after exercise (the time point of minimum mean FEV₁, when shortness of breath caused by exercise had abated), fell into two categories: symptoms specific for asthma (tight chest, wheeze, breathlessness, and cough); and non-specific symptoms (such as “out of breath”, headache, leg ache, etc). The specificity and sensitivity of any asthma specific symptom for EIA (fall in FEV₁ ⩾13%) were 82% and 50% respectively.

Confounders

None of the potential confounding factors (age, gender, asthma severity, medication, habitual exercise, attitude, and competitiveness) showed any association with the children’s perceptual scores (p = 0.2–0.9). Multiple regression analysis was therefore not carried out.

Repeatability

Both for parents and children there were great discrepancies between repeat visits, with no significant overall relation between the change in log VAS and the change in severity of EIA between visits. However, for 20/24 children the changes in VAS score and EIA were in the same direction, although 13 of these children had small differences in both. The other four children had large differences in EIA between visits, but with paradoxical changes in VAS. In one example, on visit 1, change in FEV₁ was 8% and change in log VAS was 2.3, and on visit 2, change in FEV₁ was 19%, but change in log VAS was only 0.9. These four children reported more habitual exercise than the rest of the group, but had a neutral attitude to whether exercise was healthy or fun.

DISCUSSION

This study aimed to assess the validity of a report by asthmatic children and their parents, of the severity of the child’s EIA. It showed that children’s reports of EIA following a standardised exercise test in a laboratory, relate weakly to objective measures of severity (change in FEV₁) after exercise. Children’s retrospective reports of EIA over the previous week, the kind of information which is often used in clinical management, bore no relation to the laboratory measurements. Parents’ perception of their child’s EIA, measured in the laboratory by two different scales, did not relate to any lung function measurements.

Children’s perception of EIA was not affected by any of several potential confounding factors such as age, gender, asthma severity, medication, and habitual exercise. Most of the descriptive comments made five minutes after stopping exercise (when the breathlessness of exercise had subsided and EIA was at its maximum) were unrelated to EIA. The poor outcome was not explained by the cross sectional nature of the study, since in the longitudinal part of the study, both children and parents were inconsistent in their ability to perceive symptoms on repeat visits.

Methodological issues

The two main outcome measures are the physiological measurements and the perceptual scales. FEV₁ as determined by spirometry was used as a measure of EIA. FEF_25–75 and FEF₅₀, were also measured in an attempt to identify children who may perceive changes in their small airways, but with even poorer statistical associations. Change in FEV₁ was perceived more accurately than change in PEF. Spirometry itself, as a result of forced vital capacity (“big breath”) manoeuvres, may affect airway function for up to two minutes. This is a potential problem with all such studies. In no case was there an obvious deterioration or improvement in FEV₁ within sets of repeated blows. We consider that a big breath effect was unlikely to explain the discrepancy between perception and physiology.

Exercise testing was standardised, although for the purposes of the study, this was not strictly necessary, since a wide variation of EIA was sought (and expected), both between subjects and within subjects for repeat measurements.

The VAS designed for this study was non-specific and the questions used simple terminology. The scale was well understood by children, and in fact parents had more difficulty in using the scale. Studies using specific scales have shown mixed results. Ergood and colleagues²¹ used a VAS to assess breathlessness on exercise (but not EIA), and found it correlated well with physiological measures. However, Baker and colleagues,²⁶ who measured the perception of bronchoconstriction in children after methacholine challenge with a specific breathlessness scale found that there was a poor relation with physiological measures.

The marking of a VAS can be distorted by habitual trends and patterns by individual subjects. In this study there was positive skewing towards the “no problem” end, as previously reported by others.²⁷ Individuals tend to be consistent in under- or over-rating scales. This could explain the good correlation between VAS and categorical scales. To control for this, an independent VAS enquiring about for example the perception of pain when pricked by pin, would have been helpful. However, the categorical scale was derived (for comparison) from a questionnaire previously validated by Juniper and colleagues²⁸; the VAS thus appeared to be a valid measure of perception, although neither scale correlated with physiological measures.

The study size was adequate to give a power of 80% to detect a clinically useful correlation (0.7) between perceptual score and physiological measures of EIA. However, there was a degree of self selection bias, which accounts for the uneven distribution of subject characteristics. Older children were targeted without success. Previous studies using older children have found an effect of age.^19,20 The sample also had a high proportion of mild asthma (BTS step 2), which may explain why no effect of either treatment or asthma severity was found. It was expected that children with severe asthma would have blunted perception (as has previously been shown,¹⁶ and would be expected by a simple application of Weber’s Law), but there were small numbers of children on BTS steps 3 and 4 medication. There was no effect of habitual level of exercise, attitude towards exercise, or competitiveness, which were expected to alter “norms” and hence the perception of EIA. Other confounders such as emotional state, intelligence, health beliefs, and level of bronchial hyperresponsiveness before visiting the laboratory may have altered symptom perception, but were not measured.⁸ These were, of course irrelevant in the longitudinal component of the study.

Clinical implications

This study has highlighted some important clinical issues. Health professionals need to be aware that reports of EIA may not be reliable and that children may be describing symptoms other than those of EIA. This was illustrated by the modest specificity (82%) and low sensitivity (50%) of children’s descriptions. The effect of potential confounding factors on the perception of EIA should be considered, although we failed to show their importance in this study. Parental perception of EIA was poor, and this suggests that a child’s history is more accurate. The results of this study are consistent with those of Lara and colleagues,²² who found child’s perception related to change in large airway obstruction (FEV₁ was not measured) on exercise, whereas parents’ perception did not.

Parents and children were inconsistent in their ability to perceive symptoms during repeat visits. Four of 24 children reported paradoxical changes in symptoms. If true, this has major clinical implications. The physician relies on the child and parent to be accurate and consistent at each visit in order, for example, to make decisions about the control of asthma and to modify therapy. Wilson and colleagues²⁹ showed the VAS to have poor repeatability in a time period of 2–6 weeks, but our categorical “bother” scale was equally poor. Studies in adults have shown that by using bronchial hyperresponsiveness (by methacholine challenge) as a clinical outcome target, improvements in asthma outcome can be achieved.³⁰ In children a history of EIA as a surrogate measure of bronchial hyperresponsiveness for clinical management, although widely used, appears to be less reliable. Regular lung function measurements provide an alternative, but may be no better.³¹ New techniques to target airway inflammation could be more appropriate, although they will be difficult to apply to children.³²

Further research

This study has provided some foundations for future research in this area. A step forward would involve distinguishing the level of breathlessness caused by exercise from that caused by airway obstruction; for example, on the one hand by preventing the bronchoconstriction of exercise by prior administration of a bronchodilator, and on the other by inducing airway obstruction (say, with methacholine) without exercise. There may also be a need for more qualitative research in this area, in which children describe their symptoms in terms familiar to themselves rather than by the use of scales (albeit non-specific).³⁴ Such an approach might identify more discriminating terms to describe EIA.³⁵ Finally, further information on confounding factors is required. This research would allow for further understanding of children’s concepts and illness beliefs associated with EIA and hence strengthen the value of the question “What happens to your asthma when you run around?”.