Weighting of the data and analytical approaches may account for differences in overcoming the inadequate representativeness of the respondents to the third wave of a cohort study

Anne W. Taylor, Eleonora Dal Grande, Janet Grant, Sarah Appleton, Tiffany K. Gill, Zumin Shi, Robert J. Adams

Research output: Contribution to journalArticlepeer-review

9 Citations (Scopus)


Objectives: Attrition in cohort studies can cause the data to be nonreflective of the original population. Although of little concern if intragroup comparisons are being made or cause and effect assessed, the assessment of bias was undertaken in this study so that intergroup or descriptive analyses could be undertaken. Study Design and Setting: The North West Adelaide Health Study is a chronic disease and risk factor cohort study undertaken in Adelaide, South Australia. In the original wave (1999), clinical and self-report data were collected from 4,056 adults. In the third wave (2008-2010), 2,710 adults were still actively involved. Comparisons were made against two other data sources: Australian Bureau of Statistics Estimated Residential Population and a regular conducted chronic disease and risk factor surveillance system. Results: Comparisons of demographics (age, sex, area, education, work status, and income) proved to be statistically significantly different. In addition, smoking status, body mass index, and general health status were statistically significant from the comparison group. No statistically significant differences were found for alcohol risk. Conclusion: Although the third wave of this cohort study is not representative of the broader population on the variables assessed, weighting of the data and analytical approaches can account for differences.

Original languageEnglish
Pages (from-to)461-464
Number of pages4
JournalJournal of Clinical Epidemiology
Issue number4
Publication statusPublished or Issued - Apr 2013


  • Attrition
  • Bias
  • Biomedical
  • Cohort study
  • Nonresponse
  • Representativeness

ASJC Scopus subject areas

  • Epidemiology

Cite this