Senior researcher Disasters and Environmental Hazards
Publicatie
Publication date
Self-selection in participation in the first health survey, three weeks after a man-made disaster.
Grievink, L., Velden, P.G. van der, Yzermans, C.J., Roorda, J., Stellato, R.K. Self-selection in participation in the first health survey, three weeks after a man-made disaster. European Journal of Public Health: 2005, 15(Suppl. 1), p. 130. Abstract. 13th Annual EUPHA Meeting "Promoting the public's health: reorienting health policies, linking health promotion and health care", Graz, 10-12 November 2005.
Background: Three weeks after a firework disaster in Enschede, The Netherlands, a health survey was
performed among survivors. The primary aims of the study were collecting data for health care policy
making and decreasing uncertainty concerning exposure to toxic substances. Therefore, each survivor
could participate in the study. In the aftermath of disasters selfselection is often assumed, but
the direction of the bias has rarely been studied. We were able to examine the selfselection of the
survivors and the direction of the bias. Methods: The health survey included a questionnaire on
health and exposure. This questionnaire was merged with the electronic medical records of the
survivors in general practice according to privacy directives. The GP database included most of the
survivors. Data on demographics, utilization and morbidity one year pre- and post-disaster were
used. Results: Preliminary results suggest that 26% of the affected residents (N = 1171)
participated in the health survey. This percentage was not different for the relocated and
non-relocated residents [odds ratio (OR) = 1.1, 95% CI ¼ 0.9–1.2]. Men (OR = 0.7, 95% CI ¼ 0.6–0.8)
and singles (OR = 0.5, 95% CI = 0.4–0.6) participated less often and immigrants more often (OR =
1.5, 95% CI = 1.3–1.7). Survivors who reported health complaints to their GP before and after the
disaster were more likely to participate in the health survey. For example, survivors in the health
survey presented more psychological complaints (OR ¼ 1.4, 95% CI ¼ 1.1–1.7) to their GP before the
disaster than the survivors who did not participate. Conclusions: In the survey 3 weeks after the
disaster, we had a selection of fewer men and singles and of more immigrants. Survivors with more
health complaints selected to participate, which might have lead to higher prevalence estimates of
health problems. After future disasters, information on self-selection in surveys will be necessary
to decrease the bias in the prevalence estimates on which policy is based. (aut.ref.)
performed among survivors. The primary aims of the study were collecting data for health care policy
making and decreasing uncertainty concerning exposure to toxic substances. Therefore, each survivor
could participate in the study. In the aftermath of disasters selfselection is often assumed, but
the direction of the bias has rarely been studied. We were able to examine the selfselection of the
survivors and the direction of the bias. Methods: The health survey included a questionnaire on
health and exposure. This questionnaire was merged with the electronic medical records of the
survivors in general practice according to privacy directives. The GP database included most of the
survivors. Data on demographics, utilization and morbidity one year pre- and post-disaster were
used. Results: Preliminary results suggest that 26% of the affected residents (N = 1171)
participated in the health survey. This percentage was not different for the relocated and
non-relocated residents [odds ratio (OR) = 1.1, 95% CI ¼ 0.9–1.2]. Men (OR = 0.7, 95% CI ¼ 0.6–0.8)
and singles (OR = 0.5, 95% CI = 0.4–0.6) participated less often and immigrants more often (OR =
1.5, 95% CI = 1.3–1.7). Survivors who reported health complaints to their GP before and after the
disaster were more likely to participate in the health survey. For example, survivors in the health
survey presented more psychological complaints (OR ¼ 1.4, 95% CI ¼ 1.1–1.7) to their GP before the
disaster than the survivors who did not participate. Conclusions: In the survey 3 weeks after the
disaster, we had a selection of fewer men and singles and of more immigrants. Survivors with more
health complaints selected to participate, which might have lead to higher prevalence estimates of
health problems. After future disasters, information on self-selection in surveys will be necessary
to decrease the bias in the prevalence estimates on which policy is based. (aut.ref.)
Background: Three weeks after a firework disaster in Enschede, The Netherlands, a health survey was
performed among survivors. The primary aims of the study were collecting data for health care policy
making and decreasing uncertainty concerning exposure to toxic substances. Therefore, each survivor
could participate in the study. In the aftermath of disasters selfselection is often assumed, but
the direction of the bias has rarely been studied. We were able to examine the selfselection of the
survivors and the direction of the bias. Methods: The health survey included a questionnaire on
health and exposure. This questionnaire was merged with the electronic medical records of the
survivors in general practice according to privacy directives. The GP database included most of the
survivors. Data on demographics, utilization and morbidity one year pre- and post-disaster were
used. Results: Preliminary results suggest that 26% of the affected residents (N = 1171)
participated in the health survey. This percentage was not different for the relocated and
non-relocated residents [odds ratio (OR) = 1.1, 95% CI ¼ 0.9–1.2]. Men (OR = 0.7, 95% CI ¼ 0.6–0.8)
and singles (OR = 0.5, 95% CI = 0.4–0.6) participated less often and immigrants more often (OR =
1.5, 95% CI = 1.3–1.7). Survivors who reported health complaints to their GP before and after the
disaster were more likely to participate in the health survey. For example, survivors in the health
survey presented more psychological complaints (OR ¼ 1.4, 95% CI ¼ 1.1–1.7) to their GP before the
disaster than the survivors who did not participate. Conclusions: In the survey 3 weeks after the
disaster, we had a selection of fewer men and singles and of more immigrants. Survivors with more
health complaints selected to participate, which might have lead to higher prevalence estimates of
health problems. After future disasters, information on self-selection in surveys will be necessary
to decrease the bias in the prevalence estimates on which policy is based. (aut.ref.)
performed among survivors. The primary aims of the study were collecting data for health care policy
making and decreasing uncertainty concerning exposure to toxic substances. Therefore, each survivor
could participate in the study. In the aftermath of disasters selfselection is often assumed, but
the direction of the bias has rarely been studied. We were able to examine the selfselection of the
survivors and the direction of the bias. Methods: The health survey included a questionnaire on
health and exposure. This questionnaire was merged with the electronic medical records of the
survivors in general practice according to privacy directives. The GP database included most of the
survivors. Data on demographics, utilization and morbidity one year pre- and post-disaster were
used. Results: Preliminary results suggest that 26% of the affected residents (N = 1171)
participated in the health survey. This percentage was not different for the relocated and
non-relocated residents [odds ratio (OR) = 1.1, 95% CI ¼ 0.9–1.2]. Men (OR = 0.7, 95% CI ¼ 0.6–0.8)
and singles (OR = 0.5, 95% CI = 0.4–0.6) participated less often and immigrants more often (OR =
1.5, 95% CI = 1.3–1.7). Survivors who reported health complaints to their GP before and after the
disaster were more likely to participate in the health survey. For example, survivors in the health
survey presented more psychological complaints (OR ¼ 1.4, 95% CI ¼ 1.1–1.7) to their GP before the
disaster than the survivors who did not participate. Conclusions: In the survey 3 weeks after the
disaster, we had a selection of fewer men and singles and of more immigrants. Survivors with more
health complaints selected to participate, which might have lead to higher prevalence estimates of
health problems. After future disasters, information on self-selection in surveys will be necessary
to decrease the bias in the prevalence estimates on which policy is based. (aut.ref.)
Gegevensverzameling