Annals of Epidemiology
Volume 16, Issue 10, 2006, Pages 782-788

The Importance of Estimating Selection Bias on Prevalence Estimates Shortly After a Disaster (Article)

Grievink L.* , van der Velden P.G. , Yzermans C.J. , Roorda J. , Stellato R.K.
  • a The Centre for Environmental Health Research, National Institute for Public Health and the Environment, Bilthoven, Netherlands
  • b The Centre for Environmental Health Research, National Institute for Public Health and the Environment, Bilthoven, Netherlands
  • c The Centre for Environmental Health Research, National Institute for Public Health and the Environment, Bilthoven, Netherlands
  • d The Centre for Environmental Health Research, National Institute for Public Health and the Environment, Bilthoven, Netherlands
  • e The Centre for Environmental Health Research, National Institute for Public Health and the Environment, Bilthoven, Netherlands

Abstract

Purpose: The aim was to study selective participation and its effect on prevalence estimates in a health survey of affected residents 3 weeks after a man-made disaster in The Netherlands (May 13, 2000). Methods: All affected adult residents were invited to participate. Survey (questionnaire) data were combined with electronic medical records of residents' general practitioners (GPs). Data for demographics, relocation, utilization, and morbidity 1 year predisaster and 1 year postdisaster were used. Results: The survey participation rate was 26% (N = 1171). Women (odds ratio [OR], 1.46; 95% confidence interval [CI], 1.28-1.67), those living with a partner (OR, 2.00; 95% CI, 1.72-2.33), those aged 45 to 64 years (OR, 2.00; 95% CI, 1.59-2.52), and immigrants (OR, 1.50; 95% CI, 1.30-1.74) were more likely to participate. Participation rate was not affected by relocation because of the disaster. Participants in the survey consulted their GPs for health problems in the year before and after the disaster more often than nonparticipants. Although there was selective participation, multiple imputation barely affected prevalence estimates of health problems in the survey 3 weeks postdisaster. Conclusions: Estimating actual selection bias in disaster studies gives better information about the study representativeness. This is important for policy making and providing effective health care. © 2006 Elsevier Inc. All rights reserved.

Author Keywords

Selection Bias imputation Survivors Health surveys disasters

Index Keywords

Netherlands immigrant management methodology health care policy demography selection bias human middle aged Survivors controlled study priority journal Aged general practitioner morbidity Health Surveys disaster resident Confidence interval Humans Adolescent male female electronic medical record Disasters questionnaire Physicians, Family prevalence Article health care utilization Questionnaires adult assay Delivery of Health Care health survey
Link
https://www.scopus.com/inward/record.uri?eid=2-s2.0-33749138017&doi=10.1016%2fj.annepidem.2006.04.008&partnerID=40&md5=da7ec30eb4f3a1c46c1235d7d82aa34a
DOI: 10.1016/j.annepidem.2006.04.008
ISSN: 10472797
Cited by: 42
Original Language: English