Introduction

Cough is common during childhood. It leads to many medical consultations, affects quality of life and places a considerable burden on children, families and society [1]. Cough is a non-specific symptom of various diseases. Most often it is caused by acute respiratory infections, which can make children cough 5–8 times a year for 7–9 days each time [2]. In over 90% of children, this type of acute cough resolves by the third week of the infection [3]. Some children, however, have more frequent episodes (recurrent cough, non-acute cough), and questionnaire studies have used different questions to identify these children. This can indicate an underlying severe disease, such as cystic fibrosis, or primary ciliary dyskinesia. Studies from hospital patients with cough, who had a detailed diagnostic work-up, suggest that the most frequent underlying causes for recurrent cough are asthma, protracted bacterial bronchitis, chronic ear, nose and throat diseases, gastroesophageal reflux, and prolonged bronchial responsiveness after infections such as respiratory syncytial virus, rhinovirus or pertussis (post-infectious cough) [4]. Often, however, the underlying causes are not obvious (non-specific cough) [5].

Little is known about the epidemiology of childhood cough, assessed with different standardised questions. Most studies to date are cross-sectional and based on hospital patients [4,6]. Population-based studies are scarce [7–10]. Ongoing birth-cohort studies focus on wheeze, with little attention given to cough. Prevalence of cough has usually been reported for selected age groups only [11–13]. Studies used different questions on cough: most assessed dry cough at night, a key question of the International Study of Asthma and Allergies in Childhood (ISAAC) [11,14,15]. Others used two questions from the American Thoracic Society (ATS) questionnaire: cough occurring with colds and cough apart from colds [16,17]. Some studies asked if children coughed more than their peers [18,19]. To our knowledge, it has never been investigated how prevalence of cough varies throughout childhood in the general population, and whether results depend on how the questions on cough are formulated. Host and environmental factors change as a child grows up: the immune system and the lungs develop, the time spent outdoors and physical activity pattern change, lifestyle and exposure to allergens and toxins vary. All this might affect prevalence of cough.

In a large population-based cohort, we described prevalence of cough from age 1 to 18 years, using a variety of questions to assess cough. We did this for the entire study population, and separately for boys and girls. We also assessed cough prevalence in children with and without wheeze, because cough is a key symptom of childhood asthma and we were particularly interested in children who report cough without having asthma.

Methods

The questionnaires

The questionnaire for the 1998 survey is a modified version of the questionnaire developed for the first Leicestershire cohort study, which recruited children born between 1985 and 1990 [20]. It contains questions from the ATS-childhood questionnaire [16], the key questions from ISAAC [14] and a few items from other validated questionnaires [20]. The same questionnaires were used in all subsequent surveys, with little variations to adapt the questions to growing age of children. Current wheeze (last 12 months) was assessed with the ISAAC key question [14], and wheeze was defined as ‘breathing that makes a high-pitched whistling or squeaking sound from the chest, not the throat’. We also assessed severity, whether the child ever had doctor-diagnosed asthma, and how many attacks of wheeze occurred in the past year. Repeatability of the questionnaire was good for questions on wheeze and moderate for questions on cough [21].

We asked several validated questions about cough, published in previous studies (Table 1): whether the child coughed more than other children [22]; whether he/she usually coughed with colds [16]; whether he/she coughed even without having a cold [16], and whether he/she had a dry cough at night, apart from cough associated with a cold or a chest infection [14]. We also asked about triggers for cough: triggers, related to increased breathing effort, including exercise or playing and laughing, excitement or crying; allergic triggers, such as house dust, pollen, or contact with pets; and cough triggered by food or drinks. Not all questions about cough were asked in all the surveys; the number of children who were asked a specific question is shown in supplementary S1 Table. In the first survey (1998) parents were asked whether their child had had frequent reflux during infancy. Parents completed the questionnaires until their child reached 13 years (surveys in 1998, 1999, 2001, 2003, 2006), children aged 14–17 (survey in 2010) completed the questionnaires themselves.

Download:

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

Table 1. Cough questions used in the Leicester Respiratory Cohort studies.

https://doi.org/10.1371/journal.pone.0177485.t001

Results

In 1998, most children (59%) were aged 1 year, 52% were boys, and 27% were of South Asian ethnic origin (Table 2). Prevalence of current wheeze was 34% in 1-year-olds, 23% in 2-year-olds, 19% in 3-4-year-olds, and about 15% thereafter. Prevalence of doctor-diagnosed asthma (ever in life) increased from 12% at age 1 to 23% at age 14–17 years (S2 Table). Among children with current wheeze, prevalence of doctor-diagnosed asthma increased from 33 to 64% throughout childhood, among non-wheezers from 3 to 16% (S3 Table).

Download:

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

Table 2. Characteristics of the study population in 1998 (N = 6808).

https://doi.org/10.1371/journal.pone.0177485.t002

Prevalence of cough throughout childhood; entire study population

Prevalence of cough varied significantly, depending on the questions used to assess it (Figs 1 and 2, Table 3; S4 Table). Ten percent (95%CI 10–11) of parents said that their child coughed more than peers, with little differences between age groups (Fig 1). Sixty-nine percent of children (69–70%) were reported to cough usually with colds; there were only small differences between age groups. In contrast, the prevalence of cough without colds increased with age from 34% in 1-year-olds to 55% in children aged 14–17. Night cough was reported for 25% (25–26%) of children. It increased slightly from 23% in 1-year-olds to 31% in 3-4-year-olds, and then decreased again to 20% at age 14–17.

Download:

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

Fig 1. Prevalence of coughing more than peers, cough with colds, cough apart from colds and night cough in the entire cohort, and among children with wheeze and non-wheezers, for different age groups.

Prevalence (%) with 95% confidence intervals.

https://doi.org/10.1371/journal.pone.0177485.g001

Download:

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

Fig 2. Prevalence of cough triggered by exercise, house dust, pollen, pets, food and laughter or crying in the entire cohort, and among children with wheeze and non-wheezers, for different age groups.

Prevalence (%) with 95% confidence intervals.

https://doi.org/10.1371/journal.pone.0177485.g002

Download:

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

Table 3. Prevalence of cough in different age groups in the entire cohort, and separately for children with and without wheeze.

https://doi.org/10.1371/journal.pone.0177485.t003

Replies to questions on triggers for cough varied strongly with age (Fig 2, Table 3; S4 Table). Ten percent of 1-year-olds were reported to cough after exercise or playing; this increased to 26% in the oldest age group. About a quarter of children in all age groups (range 18–26%) were reported to cough after laughing or crying. The prevalence of cough triggered by house dust increased from 2% in the youngest children to 14% in the oldest. Cough triggered by pollen increased from 10% in 3-4-year-olds to 15% in 7-9-year-olds, with little change thereafter. Cough triggered by pets was rare (2–4% of children) in all age groups. The proportion of families owning a pet ranged between 41 and 51% across childhood. The prevalence of cough triggered by food or drinks decreased slowly from 11% in the youngest age group to 6% in 7-9-year-olds, with little change thereafter. There was a significant association between reports of cough triggered by food and frequent reflux in infancy; 18% of 1-4-year-old children with frequent reflux were reported to have cough triggered by food compared to 6% of children without (p<0.001) (S5 Table). Children whose cough was triggered by food, had around twice as high prevalence of night cough, compared to children without cough triggered by food throughout childhood (p<0.001).

Discussion

This study describes cough prevalence systematically from infancy through adolescence in an unselected childhood cohort, using a series of questions to assess cough. We found that: 1) prevalence estimates of cough varied widely depending on the questions assessing it; 2) there were distinct age-related changes in prevalence, which differed between cough questions; and 3) all types of cough were more common in children with wheeze, but age-related prevalence patterns were similar in children with and without wheeze.

To our knowledge, ours is the first study that describes prevalence of cough, assessed by multiple questions, in a population-based sample of children covering all age groups from infancy through adolescence and with prevalences stratified by sex and by presence of wheeze. Its strengths include the large size, the fact that it is based on an unselected sample of children from the general population, that it included children of all ages (1–18 years), and used many questions to inquire about cough. Surveys were conducted every 2–3 years, always in late spring (May), used a consistent methodology, and repeated the same questions. The repeatability of the cough questions was assessed for 1-year-olds by resending the questionnaires after 3 months (August). The kappa coefficients were 0.57 for coughing more than peers, 0.46 and 0.54 for coughing with and without colds, 0.39 for night cough, and 0.53, 0.57, 0.61, and 0.49 for cough triggered by exercise, pets, food and laughter (moderate repeatability) [21]. This reflects seasonal variation in cough prevalence. It should not have biased the comparison between age groups, since all surveys had been conducted in the same season. Because there was attrition in the cohort, and we did not know if responders might differ from non-responders in cough prevalence, we performed a sensitivity analysis with appropriate weights, adjusting for potential non-response bias. These prevalence estimates were very similar to estimates in the original unweighted analysis, which suggests our results are representative of the general population. A weakness of the study is that we had not assessed measures of severity of cough, for instance, duration of episodes or sputum production. Questions on the duration of cough episodes were only included in the 2010 questionnaire, but not in the others, and could thus not be described for all age groups. This study also did not describe specific diseases leading to cough, such as whooping cough, and did not clearly differentiate between wet and dry cough. Parental or self-reporting of respiratory symptoms is a pragmatic tool in large epidemiological studies. However, it has been shown that parental reporting of cough is not very accurate compared to objective measurements [24,25]. We cannot exclude a differential recall bias, in that parents of children with wheeze might be alert to hearing and reporting night-time cough, because this is a question that doctors ask them at follow-up visits. Additionally, though symptoms from age 1–13 were parent-reported, symptoms in 14-17-year-olds were self-reported. We therefore cannot say whether changes in prevalence between the age groups 10–13 and 14–17 reflect a biological change, related to age, or a reporting difference between children and their parents [26].

Surprisingly few studies have assessed prevalence of cough in the general population (Table 4). Most studies asked only one or two questions about cough, usually the ISAAC core question on night cough, and investigated one or two age groups. Many studies described night cough in 6-7-year-olds and 13-14-year-olds as part of the ISAAC international surveys [11]. Pearce et al. compared the prevalence of asthma symptoms worldwide in almost 500,000 children [11]. Night-cough prevalence showed significant geographical variability in parallel with wheeze, with higher prevalence in English-language countries and Western Europe, and lower prevalence in Eastern Europe. Results from the UK were similar to ours: 27% for 6–7 and 19% for 13-14-year-olds. Two other British studies found slightly higher prevalence of night cough than our study [12,15], but one was conducted in patients from a deprived suburb area, registered in a general practice [12]. Robertson et al. examined night cough in almost 11,000 children from Australia, aged 7 (cough prevalence 28%), 12 (19%) and 15 years (16%), with similar findings to ours [22,27]. Night-cough prevalence in France [13], Germany [28,29], Switzerland [27,30], the Netherlands [31,32], Eastern and Northern Europe [33,34] was lower, and in Chile [27] higher, than what we found. This parallels known geographical differences for prevalence of wheeze and asthma [11].

Download:

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

Table 4. International comparison of age-specific prevalence of cough, assessed by different questions.

https://doi.org/10.1371/journal.pone.0177485.t004

The other cough questions have hardly been studied in unselected samples of children so far. Two older surveys, the first Leicester Respiratory Cohort study conducted in children born 1985–1990 [17], and a study by Burr et al. [15], reported lower prevalence of cough without colds than we found in the second Leicester cohort. This mirrors secular changes in prevalence of wheeze in the Leicestershire region [35].

The prevalence of cough varied minimally by sex, but in the same direction as the variations described for wheeze [36], which is usually more frequent in boys during preschool and early school years, but more frequent in girls from about 16 years onwards [37]. In our study, we found that prevalence of cough was higher in boys until the age of about 13 but higher in girls thereafter—as also found in other studies for selected questions [15,37]. The gender difference was most pronounced for types of cough, which have been reported to be related to asthma (cough more than peers, cough without colds, cough at night, cough triggered by exercise or certain aeroallergens). Hence, some of the underlying mechanisms for gender differences might be similar for cough and wheeze. Bronchial hyperresponsiveness in childhood is more common and more severe among boys, but girls catch up during adolescence [36]. Second, sex-based differences in the anatomy of the respiratory system have been described as dysanapsis—boys have narrower airways related to the lung volume, caused by disproportionate growth of airways and lung parenchyma [38]. The reverse is seen in adults where airway diameters are about 17% larger in men [39]. A third mechanism could be age-related changes in cough sensitivity: some studies in adults showed higher cough sensitivity in women [40–42], while Chang et al. found no sex differences in cough receptor sensitivity in young children [43,44].

Only few studies measured cough prevalence separately in children with and without wheeze. We confirmed that the cough prevalence is much higher in children who wheeze than in non-wheezers. Brooke et al. has previously shown that night-cough prevalence was 39% in 4-8-year-olds with wheeze, and 19% in those without [25]. Von Mutius et al. assessed exercise-induced cough with a slightly different question, and reported that 15% of 9-11-year-olds overall, and 51% of those with asthma, had exercise-induced cough. Results for children without wheeze were not given [45]. Night-time cough and cough without colds are thought to be key symptoms of childhood asthma and are usually asked in consultations. Surprisingly, only a quarter of all children, and only 46% of children who have wheeze, complained about night cough. The proportion increased in children with frequent wheeze, or those with doctor-diagnosed asthma, but was not higher than 67% for night cough or than 81% for cough apart from colds. This could reflect differences in importance and occurrence of cough between different asthma phenotypes, as found in a previous study that considered chronic cough in a multivariate model to define asthma phenotypes [46].

Overall, prevalence of cough varies throughout childhood in distinctive age-related patterns, and results differ depending on which exact question is used to assess cough. About 10% of parents, independent of their child’s age, think that their child coughs more than others; about 30% of parents whose children wheeze, but only about 5% of parents whose children do not wheeze. This suggests that the burden of recurrent cough in children without wheeze is relatively low. Unexpectedly, only about 70% of children report coughing with colds. This could signal differences in cough receptor sensitivity, as children may have variable thresholds for reflex initiation [47]. Cough without colds gets more common with age. Reasons might include cough caused by smoking, or because of undiagnosed asthma [48]. Asthma is often underdiagnosed in adolescents, and cough is the most common symptom among adolescents with undiagnosed asthma [49].

Prevalence of cough triggered by different factors varied strongly with age. Our data do not allow making any deduction on the underlying causes of cough. However, based on other literature, we can make some speculations, which need to be verified by further research. Cough triggered by exercise was more frequent in teenagers than in younger children, probably because teenagers are more physically active or perform more strenuous activities. Cough triggered by laughter or crying remained comparatively constant throughout childhood. Its pathophysiological mechanisms are probably similar to exercise-induced cough (increased breathing effort, dry airways, consequently cough) [50]. It has been suggested that these children have low threshold mechanosensors, which render even mild mechanical triggers irritating, possibly as part of cough hypersensitivity syndrome [47]. Prevalence of coughing due to allergic triggers (house dust, pollen and pet dander) increased with age. We know that allergic sensitisation to allergens increases with age and some monosensitised children become polysensitised [51]. We found similar prevalence of food-induced cough in small children, whether they had current wheeze or not. This suggests that the causes of cough triggered by food might vary by age. In infants, the main cause is likely to be gastroesophageal reflux [52]. In fact, cough triggered by food was reported more often in young children with reflux (17–20%) compared to those without (5–7%). In toddlers, food allergies might become more important, as reflected in our study; children with wheeze had more food-induced cough and the cough prevalence diverged between the presence and absence of wheeze in older children. There was more night-time cough among children who reported cough triggered by food compared to children without, suggesting that a proportion of these infants aspirate when they are in supine position at night.

In summary, our study describes for the first time prevalence of cough throughout the whole childhood, with a standardised approach and a wide variety of questions to assess cough. We found large differences between different cough questions, implying that the choice of questions relating to cough and their phrasing are essential, and comparisons between studies are only valid for similar questions and age groups. Future studies should select the questions on cough carefully, depending on the exact aim of the study and age of participants. We found differences between girls and boys, and differences in age-related cough patterns, with older children more commonly reporting exercise-induced cough, and cough in conjunction with aeroallergen sensitisation. The age-related patterns are likely explained by physiological changes in the growing child and changing exposures to environmental factors.

Supporting information

Acknowledgments

We sincerely thank the cohort participants and the parents of the Leicester Respiratory Cohorts for completing the questionnaires. We thank Dr. Anina M. Pescatore and Florian S. Halbeisen for their help with sensitivity analysis and contribution with S1 Figure. We thank Dr. Kali Tal for her editorial assistance.