Population Characteristics

We combined data from interviews, physical examinations, and accelerometry from two cohorts of German Caucasians: GINIPlus and LISAPlus, born between 1995 and 1999. Further details on study designs of GINIplus [25, 26] and LISAplus [27] are published elsewhere. Both studies were approved by the respective local Ethics Committees (Bavarian General Medical Council, Medical Council of North-Rhine-Westphalia) and by written consent from participating families.

GINIPlus (German Infant Nutritional Intervention PLUS environmental and genetic influences on allergy development) was initiated to investigate allergy development. Of 5991 infants recruited at birth, 2252 had a family history of atopy and thus were given hydrolysed baby formulas. The remainder was given no formula. At age 15, 3199 adolescents were recontacted and approached for accelerometry. 1890 subjects (59%) consented to accelerometry, of whom 1290 (40%) successfully completed and 1054 (33%) passed quality control. Ultimately 847 of these (26%) were included in the present study. For further details on recruitment, formulas, and followup see von Berg et al. (2015) [28]and Smith et al. (2016) [29]; for details on accelerometry response see Smith et al. (2016.) [23]

LISAPlus (Lifestyle-Immune-System-Allergy; Influence of lifestyle factors on the development of the immune system and allergies plus the influence of traffic emissions and genetics) is a population-based cohort of 3097 unselected infants[30] from the cities of Munich, Wesel, Bad Honnef and Leipzig. 1534 subjects were followed up at age 15, of which 1107 (64%) were from Munich or Wesel and thus approached for accelerometry. 655 subjects (59%) consented to accelerometry, of whom 435 (39%) successfully completed and 357 (32%) passed quality control. Ultimately 290 (24%) were included in the current study. [23]

For a flowchart on accelerometry recruitment, response, and completion see Smith et al. (2016.) [23] Of 1411 subjects who completed accelerometry, [23] 1137 (83%) had complete data and were included in the current paper.

Measurements of PA

Accelerometry was combined with an activity diary to document activity domains and weartime. Detailed descriptions of accelerometer protocol, quality control, and data cleaning are given elsewhere. [31, 32] [23] Accelerometers (ActiGraph GT3X, Pensacola, Florida) were worn at the hip. Sampling rate was 30 Hz; accelerations were stored at 1 Hz and converted into activity levels in one-minute epochs using the algorithm for children from Freedson et al, 2005.[33] Diaried weartime was validated against that indicated by the device according to the algorithm of Troiano et al. (2007) [34], using SAS programs from NHANES[31]. Only data from waking time during diaried and validated monitor wear were used. For more details on accelerometry protocol see Smith et al, [23].

Activity diaries contained lines for time of getting up and going to bed, participation in sport, time and reason of removing the monitor, and other activities in a standardized diary, particularly leisure time sport duration and type. Valid days had at least 10 hours of valid recording, or 7 if the subject was awake for between 7 and 10 hours. Valid subjects provided at least 3 valid weekdays, and one valid weekend day. To profile daily activity, we considered average daily minutes of moderate and vigorous PA (MPA, VPA), and sport participation.

Sociodemographic and Anthropometric Confounders

All multivariable models were corrected for correlates of PA not of primary interest. These were age, height, study center (Munich or Wesel), season of accelerometry, and nutritional intervention. For details see S1 File. Initial analyses suggested sex-specific results, so all presented models are stratified by sex.

Allergic Respiratory Conditions

Allergic respiratory conditions were asthma, allergic rhinitis and atopic sensitization to aero-allergens, defined as follows:

• Asthma: As in Mölter et al. (2015) [18] and Jarvis et al (2012)[35] current asthma at 15 years was defined as having at least two of the following: doctor diagnosis of asthma ever between age 3–15, current wheezing at 15 years of age, and asthma medication at 15 years of age.
• Allergic rhinitis: Allergic rhinitis was defined as a doctor diagnosis of either allergic rhinitis or hayfever at any time in the past year. Adolescents with asthma or asthma medication were treated in the models as asthma and excluded from the rhinitis group, even if they also had rhinitis.
• Aero-allergen sensitization (atopy): Atopic sensitization was defined as any sensitization to aero-allergens compared with none, defined as at least one RAST positive (IgE ≥0.35 kU/l) for the following airborne allergens: birch, mugwort, ambrosia, grass, rye, dogs, cats, dust mites (Dermatophagoides pteronyssinus) and indoor mold (Cladosporium herbarum). Models of atopy excluded children with current rhinitis or asthma, or current medication for rhinitis or asthma.
• Lung-healthy (control): Lung-health in this population was defined as: since the age of 3 years no asthma, and in the past year no asthma symptoms (wheezing), no rhinitis, no rhinitis symptoms (nose/eye symptoms; see below), no asthma or rhinitis medications, no positive aero-allergen RAST, or positive bronchodilator response (see Supplement and Miller et al.(2005) [36]for more detail on testing)

Allergic Respiratory Symptoms

• Wheezing: Self-reported wheezing, whistling or chest tightness in the past year was included as an indicator of incompletely diagnosed or treated respiratory disease, particularly asthma, which may discourage vigorous PA.
• Nose and eye symptoms: Self-reported current nose and eye symptoms (runny nose, itchy eyes) in the past 12 months in the absence of a cold were included as an indicator of possible untreated or incompletely treated rhinitis, which may in turn discourage PA.

Statistical Methods

All calculations were done using SAS 9.2 or 9.3 (Cary, NC.)

Populations (Tables 1 and 2) were compared using nonparametric tests: Kruskal-Wallis for multilevel categorical variables, Wilcoxon’s two-tailed rank-sum test for all others. To model corrected relationships, sex-stratified generalized linear models were used to model PA outcomes as statistical functions of respiratory diseases and conditions, corrected for age, height, study center, nutritional intervention, season of accelerometry (categorical) and parental education.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Population and Selection.

https://doi.org/10.1371/journal.pone.0161461.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Population and Selection.

Comparison between lung-healthy controls and children with respiratory allergies.

https://doi.org/10.1371/journal.pone.0161461.t002

Each PA measure (MPA, VPA, and sport participation) was modeled as function of all confounders not of primary interest, as well as one respiratory condition or diagnosis at a time. Odds of any sport participation during accelerometry were modelled logistically and are presented as odds ratios. Daily minutes MPA and VPA were log-transformed for normality before modeling, and inspection of histograms and q-q plots confirmed normality. Results are presented as percent difference from children without the symptom (in models of symptoms) or as percent difference from apparently lung-healthy controls (in models of asthma, rhinitis, and atopy.)

Inclusion Criteria

To separate the effects of the different allergic diagnoses, these conditions were modeled separately[37] similar to the comparisons made in Mitchell et al, 2013.[4] Children who confirmed asthma (n = 94) were one group; of the remainder, those who confirmed allergic rhinitis were another group (n = 90); of the remainder, those with positive aero-allergen RAST were the third group (n = 363). Children with each condition were compared only to lung-healthy controls (detailed above; n = 590.) However, all subjects were included in the models of symptoms.

Children who did not confirm any diagnosis, but who also did not fit our criteria for lung-health, were excluded from the analysis. For example, this included children who did not confirm asthma or allergic rhinitis and who did not have a positive RAST, but had a positive bronchodilator response (n = 41) and/or asthma in childhood (n = 49) medications for asthma (n = 2) or rhinitis (n = 5), nose/eye symptoms (n = 51) or wheezing (n = 25.) Ultimately of 1411 subjects who completed accelerometry, 1137 (81%) were included in the paper.

Main Findings

In this large study of European adolescents we found that objectively-measured physical activity, particularly vigorous activity, was lower in adolescent boys with asthma and/or rhinitis, who nevertheless participated fully in sport; that symptoms did not appear to constrain PA; and that there were no effects in girls.

Interpretation in Relation to Previous Work

Like others, [6]^,[16] we found sex-specific associations between asthma, rhinitis and PA. However, while Groth et al. (2016) [6] and the current study found that asthma was associated with PA only in boys, Yiallouros et al. (2015)[16] found an effect only in girls, while Sugimoto et al[5] found that rhinitis was associated with lower PA only in boys. This heterogeneity may be explained by the known gender differences in prevalence, symptoms, and progression of asthma [39–42] and/or cultural differences in acceptability of sport [43, 44] especially for girls or children with asthma. [2] Furthermore, asthma diagnosis and treatment may be more available in our own population of Germans than in either the low-income Americans studied in Groth et al. (2016) [6] and Firrincieli et al. (2005)[13], or the Cypriots studied in Yiallouros et al. (2015)[16]. Lastly, our population was 2 years older than those in Groth et al. (2016)[6] and 7 years older than those in Yiallouros et al. (2015)[16] and age is known to impact on both activity pattern[45] and allergy development. [42]

We concur with the literature [2–4] that allergic phenotype, especially in boys, is often, but not always, associated with low PA; and find that neither respiratory symptoms[4] nor avoidance of sport[17] necessarily drive this difference. Direct comparison of effect sizes is problematic since some studies quantify PA as MVPA, some as VPA only, and some as sport participation; and at least in our cohort these measures of PA had different relationships with allergy. Furthermore many studies, including our own, are near the detection limit at p = 0.05 and thus are subject to effect inflation by publication bias. However, children with asthma or asthma symptoms typically got 30–50% less PA or sport in a recent review by Williams et al.[2] Asthma was associated with 30% less self-reported VPA and MPA in boys by Groth et al. (2016)[6] and 41% less accelerometric MVPA in girls by Yiallouros et al. (2015)[16] who like us found no effect of wheezing. Thus, our findings in boys are comparable[6] or at the lower end[2] of reported effects.

Intercomparison of studies is further hampered by the tendency to compare populations without correction for confounding, either by conditions such as overweight or by intercorrelations between allergic conditions. For example, Sugimoto et al. (2012) [5] found a 30% decrease in VPA among boys with rhinitis, but 35% of children with rhinitis also had asthma; and although rhinitis was not independently associated with VPA in our cohort, asthma was. Such comorbidity may have driven the association found by Sugimoto et al. (2012).

The lowered VPA in boys with asthma is consistent with the hypothesis that allergic symptoms or exercise-induced bronchoconstriction (EIB) may constrain PA. Wheezing was associated with 42% less accelerometric PA in Firrincieli et al. (2005)[13] and 30% less self-reported VPA in Groth et al. (2016.) [6] Despite a high percentage of medication use in our population, symptoms of asthma and rhinitis remained prevalent. Although we did not inquire whether it occurred specifically during exercise, wheezing is a common symptom of incompletely controlled asthma, typically occuring during VPA and/or cold weather.[18] In line with this observation, we found that in boys with asthma MPA was almost normal but VPA was much more significantly reduced, an observation also reported by Weisgerber et al. (2008) [46]. Likewise, Sugimoto et al. (2012)[5] found that in a population with a 35% prevalence of asthma MPA was almost normal, but VPA was obviously decreased.

Unlike others, we found that sport participation was not associated with any allergic diagnosis or symptom in either sex. However, we observed high rates of sport participation in general (over 2/3 of children) and previous research with this cohort [22, 47] found that neither female gender nor overweight was associated with sport. When combined with the fact that school sport is mandatory in Germany, it appears that sport is generally well accepted within this population and that barriers to participation are low. This was not the case in British populations studied by Williams et al. (2008), where asthmatic children avoid sport,[2] or are prevented from participating, [2] for fear of exacerbations.

Together, our data show less PA in boys with asthma and/or rhinitis even in the presence of comparable sport participation. Clinicians should fully control asthma and rhinitis in children in order to enable them to participate fully in VPA; and counsel these children and their parents of the possibility, demonstrated in this paper, of full participation in PA and sport. Activity-induced symptoms and consequent limitation of PA should be viewed as indicators to improve the treatment, ensure medication intake as prescribed, and thus eliminate the excuse for inactivity.

Strengths and Limitations

Our subjects are a generally healthy, prosperous, and uniform group, recruited as follow-up from two birth cohorts and further selected for completion of interviews, a physical exam, wearing of an accelerometer, and meticulous keeping of diaries. Thus these adolescents are likely more compliant and thorough than their peers. They also live in a culture where PA and sport are common: sport participation levels are generally high, and we did not observe any tendency for lower participation in allergic adolescents or those with allergic parents (not shown.) This is known to not be the case elsewhere,[10] and relationships we find are thus necessarily population-specific.

However, since allergic symptoms and doctor diagnoses were reported by either participants or their parents ascertainment bias is a concern. Health-conscious individuals may recognize symptoms or remember diagnoses more accurately, or seek treatment more aggressively: however, relationships were stable in a sensitivity analysis limited to adolescents with negative bronchodilator response (not shown) suggesting that ascertainment bias did not drive results. Likewise, accelerometric data represent a snapshot of activity over a period of a week or less. While we made efforts to include only representative days during school time and to correct for seasonal effects, it is likely that some bias remains.

Lastly, sample size is a concern. Effect inflation by publication bias has been addressed above; to which we add that although our study population was large, it was mostly healthy. We sampled less than 100 adolescents with asthma and found an effect in only half of them (boys.) While no single outlier of either sex drove our results (not shown) and our model treated asthma and rhinitis as mutually exclusive (i.e. the model of rhinitis excluded adolescents with asthma) MPA and VPA were not. Because MPA and VPA tended to intercorrelate within subject (not shown) it is somewhat to be expected that a group of adolescents with low VPA (boys with asthma) also had low MPA.

Implications for Future Research, Policy and Practice

Our research shows that at least some aspects of the relationship between PA and allergic phenotype are population-specific. Many previously found associations, particularly those between symptoms and PA and between sport and allergy, were not found in our population and thus may be modifiable. However, we confirm that asthma and rhinitis in boys are associated with low PA independent of each other and of reported symptoms, with effects most severe for asthma. We also find that even when allergic symptoms are present, allergic diagnoses do not necessarily preclude adolescents’ sport participation. Future research should further explore these links rather than taking them as proven.

Low PA by allergic children is in and of itself a health risk. Coaches, clinicians, and patients may need to collaborate to ensure adequate treatment of allergy and full participation in both sport and PA. To ensure equal access to the health benefits of PA, it may be necessary to target interventions or education to allergic children and/or their parents. PA performance should be specifically addressed by the treating physician, and limitations due to inadequate control of asthma and/or rhinitis should cause either improved treatment schemes and/or regular medication use by the child. Furthermore, counseling as to the need for PA should perhaps target the parents of allergic children, as well as the children themselves.

More generally, the risk of low PA was elevated in girls as compared to boys. Although asthmatic boys’ VPA was significantly lower than that of lung-healthy boys, it was almost the same as that of lung-healthy girls. This confirms previous research by us [23]and by others[44] [20] which shows that males are significantly more active than females. Since recommended PA levels are the same for both sexes, [48] this difference suggests that girls are particularly vulnerable to inactivity, and thus perhaps at greater risk for inactivity-related diseases. Future research and interventions should focus on increasing PA in girls.