Non-response did not affect prevalence estimates of asthma and respiratory symptoms - results from a postal questionnaire survey of the general population

Abstract

Background: A high participation rate is warranted in order to ensure validity in surveys of the general population. However, participation rates in such studies have declined during the last decades.

Objective: To evaluate the reasons for and potential effects of non-response in a large population-based survey about asthma and respiratory symptoms in Northern Sweden.

Methods: Within the Obstructive Lung Disease In Norrbotten (OLIN) studies, a random sample of 12,000 adults aged 20-79 was invited to a postal questionnaire survey about asthma, allergic rhino-conjunctivitis and respiratory symptoms in 2016. Three reminders were sent. A random sample of 500 non-responders was invited to a telephone interview.

Results: The participation rate in the initial mailing was 41.4%, and 9.2%, 5.0%, and 2.6% in the subsequent three reminders and totally 58.3% (n = 6854) responded. Of 500 non-responders selected for telephone interviews, 320 were possible to reach and 272 participated. Male sex, younger age, and current smoking were associated with both late and non-response. The prevalence of asthma and most respiratory symptoms did not differ significantly between responders and non-responders while allergic rhino-conjunctivitis and smoking was more common among non-responders. Reminders increased the participation rate but did not alter risk ratios for smoking and occupational exposures. Reasons for non-response were mainly lack of time and having forgotten to answer.

Conclusions: With a response rate of 58.3%, neither the prevalence estimates of asthma, respiratory symptoms nor the associations to risk factors were affected by non-response, while allergic rhino-conjunctivitis and smoking was underestimated in this Swedish population.

Keywords: Non-response; Participation rate; Population survey; Respiratory epidemiology.