Participation bias in a randomised trial of screening for lung cancer

Lung Cancer. 2011 Sep;73(3):325-31. doi: 10.1016/j.lungcan.2010.12.018. Epub 2011 Feb 15.

Abstract

Introduction: Participation bias might affect the results and the representability of randomised controlled trials. We investigated the degree of socio-demographic and psychosocial participation bias in the Danish Lung Cancer Screening Trial (DLCST).

Methods: In DLCST the questionnaire COS-LC (Consequences of Screening in Lung Cancer) was used to measure the psychosocial aspects of screening. To investigate a difference with a comparable representative sample from the Danish population, we sent out an inclusion questionnaire to 3999 Danes in the age from 50 to 70 years randomly selected from the Central National Register. Those who completed the inclusion questionnaire and met the inclusion criteria from DLCST received the COS-LC. Those who completed the COS-LC-and thus formed a population sample comparable to DLCST - were compared to the DLCST participants on socio-demographics and psychosocial measures.

Results: Participation rates were high among the comparable population sample: 75.3% completed the inclusion questionnaire and 77.4% of those who were eligible completed the COS-LC. The analyses revealed differences between the DLCST participants and the comparable population sample in the following socio-demographic aspects: social group, living alone, gender, age and geographical area. DLCST participants reported less negative psychosocial aspects than the comparable population sample.

Conclusion: The present study has shown substantial socio-demographic and psychosocial participation bias in DLCST.

Publication types

  • Comparative Study
  • Randomized Controlled Trial

MeSH terms

  • Aged
  • Denmark
  • Early Detection of Cancer / statistics & numerical data
  • Female
  • Humans
  • Lung Neoplasms / diagnosis
  • Lung Neoplasms / epidemiology*
  • Lung Neoplasms / psychology
  • Male
  • Middle Aged
  • Patient Selection
  • Population Groups / statistics & numerical data*
  • Psychometrics / statistics & numerical data
  • Randomized Controlled Trials as Topic
  • Selection Bias
  • Socioeconomic Factors*
  • Surveys and Questionnaires