Prevalence of childhood cough in epidemiological studies depends on the question used: findings from two population-based studies

Abstract

Background: Epidemiological studies use different questions to assess recurrent cough in children. In two independent population-based studies, we assessed how prevalence estimates of cough vary depending on the questions parents are asked about their child's cough and how answers to the different questions overlap.

Methods: We analysed cross-sectional data from two population-based studies on respiratory health: LuftiBus in the School (LUIS), conducted in 2013-2016 among 6- to 17-year-school children in the Canton of Zurich, Switzerland, and the 1998 Leicester Respiratory Cohort (LRC) study, UK where we used data from 6- to 8-year-old children from the 2003 follow-up survey. Both studies used parental questionnaires that included the same three questions on the child's cough, namely cough without a cold, dry cough at night and coughing more than others. We assessed how the prevalence of cough varied depending on the question and how answers to the different questions on cough overlapped. We also assessed how results were influenced by age, sex, presence of wheeze and parental education.

Results: We included 3457 children aged 6-17 years from LUIS and 2100 children aged 6-8 years from LRC. All respiratory outcomes - cough, wheeze and physician-diagnosed asthma - were reported twice as often in the LRC as in LUIS. We found large differences in the prevalence of parent-reported cough between the three cough questions. In LUIS, 880 (25%) parents reported cough without a cold, 394 (11%) dry night cough, and 159 (5%) reported that their child coughed more than other children. In the LRC, these numbers were 1003 (48%), 527 (25%) and 227 (11%). There was only partial overlap of answers, with 89 (3%) answering yes to all questions in LUIS and 168 (8%) in LRC. Prevalence of all types of cough and overlap between the cough questions was higher in children with current wheeze.

Conclusion: In both population-based studies prevalence estimates of cough depended strongly on the question used to assess cough with only partial overlap of responses to different questions. Epidemiological studies on cough can only be compared if they used exactly the same questions for cough.