Validity of family history data on PD: Evidence for a family information bias

A. Elbaz, S. K. McDonnell, D. M. Maraganore, K. J. Strain, Daniel J Schaid, James Howard Bower, J. E. Ahlskog, Walter A Rocca

Research output: Contribution to journalArticle

45 Citations (Scopus)

Abstract

Objective: To study the validity of information provided by case and control subjects (or their proxies) about PD among their first-degree relatives. Methods: Secondary cases of PD were assessed both through a single informant (family history method) and through the study of each relative (family study method). The family study method was considered as the standard for comparison, and the sensitivity and specificity of the family history method were studied. Results: A total of 133 population-based case subjects and their 655 relatives were recruited, and 119 population-based control subjects and their 511 relatives. Sensitivity was 68% (95% CI = 47 to 85) for cases and 45% (95% CI = 17 to 77) for controls. Specificity was 99% (95% CI = 98 to 99) for cases and 100% (95% CI = 99 to 100) for controls. The odds ratio (OR) for family history of PD was 4.34 (95% CI = 1.63 to 11.58, p = 0.003) using the family history method and 1.86 (95% CI = 0.78 to 4.44, p = 0.16) using the family study method. The former significant OR more than doubled the latter not significant OR (relative bias = 133%). Bias was more pronounced for proxy interviews and for women informants, and when the relatives were siblings, were living, and were examined or had medical record documentation. Conclusions: Case subjects with PD (or their proxies) are more aware of PD among their first-degree relatives than control subjects (or their proxies); however, they overreport PD in relatives who are not affected. This causes a substantial family information bias.

Original languageEnglish (US)
Pages (from-to)11-17
Number of pages7
JournalNeurology
Volume61
Issue number1
StatePublished - Jul 8 2003

Fingerprint

Proxy
Odds Ratio
Population Control
Documentation
Medical Records
Siblings
Interviews
Sensitivity and Specificity
Population

ASJC Scopus subject areas

  • Neuroscience(all)

Cite this

Validity of family history data on PD : Evidence for a family information bias. / Elbaz, A.; McDonnell, S. K.; Maraganore, D. M.; Strain, K. J.; Schaid, Daniel J; Bower, James Howard; Ahlskog, J. E.; Rocca, Walter A.

In: Neurology, Vol. 61, No. 1, 08.07.2003, p. 11-17.

Research output: Contribution to journalArticle

Elbaz, A, McDonnell, SK, Maraganore, DM, Strain, KJ, Schaid, DJ, Bower, JH, Ahlskog, JE & Rocca, WA 2003, 'Validity of family history data on PD: Evidence for a family information bias', Neurology, vol. 61, no. 1, pp. 11-17.
Elbaz A, McDonnell SK, Maraganore DM, Strain KJ, Schaid DJ, Bower JH et al. Validity of family history data on PD: Evidence for a family information bias. Neurology. 2003 Jul 8;61(1):11-17.
Elbaz, A. ; McDonnell, S. K. ; Maraganore, D. M. ; Strain, K. J. ; Schaid, Daniel J ; Bower, James Howard ; Ahlskog, J. E. ; Rocca, Walter A. / Validity of family history data on PD : Evidence for a family information bias. In: Neurology. 2003 ; Vol. 61, No. 1. pp. 11-17.
@article{a224ce16f2f84ab28569c6d15898da30,
title = "Validity of family history data on PD: Evidence for a family information bias",
abstract = "Objective: To study the validity of information provided by case and control subjects (or their proxies) about PD among their first-degree relatives. Methods: Secondary cases of PD were assessed both through a single informant (family history method) and through the study of each relative (family study method). The family study method was considered as the standard for comparison, and the sensitivity and specificity of the family history method were studied. Results: A total of 133 population-based case subjects and their 655 relatives were recruited, and 119 population-based control subjects and their 511 relatives. Sensitivity was 68{\%} (95{\%} CI = 47 to 85) for cases and 45{\%} (95{\%} CI = 17 to 77) for controls. Specificity was 99{\%} (95{\%} CI = 98 to 99) for cases and 100{\%} (95{\%} CI = 99 to 100) for controls. The odds ratio (OR) for family history of PD was 4.34 (95{\%} CI = 1.63 to 11.58, p = 0.003) using the family history method and 1.86 (95{\%} CI = 0.78 to 4.44, p = 0.16) using the family study method. The former significant OR more than doubled the latter not significant OR (relative bias = 133{\%}). Bias was more pronounced for proxy interviews and for women informants, and when the relatives were siblings, were living, and were examined or had medical record documentation. Conclusions: Case subjects with PD (or their proxies) are more aware of PD among their first-degree relatives than control subjects (or their proxies); however, they overreport PD in relatives who are not affected. This causes a substantial family information bias.",
author = "A. Elbaz and McDonnell, {S. K.} and Maraganore, {D. M.} and Strain, {K. J.} and Schaid, {Daniel J} and Bower, {James Howard} and Ahlskog, {J. E.} and Rocca, {Walter A}",
year = "2003",
month = "7",
day = "8",
language = "English (US)",
volume = "61",
pages = "11--17",
journal = "Neurology",
issn = "0028-3878",
publisher = "Lippincott Williams and Wilkins",
number = "1",

}

TY - JOUR

T1 - Validity of family history data on PD

T2 - Evidence for a family information bias

AU - Elbaz, A.

AU - McDonnell, S. K.

AU - Maraganore, D. M.

AU - Strain, K. J.

AU - Schaid, Daniel J

AU - Bower, James Howard

AU - Ahlskog, J. E.

AU - Rocca, Walter A

PY - 2003/7/8

Y1 - 2003/7/8

N2 - Objective: To study the validity of information provided by case and control subjects (or their proxies) about PD among their first-degree relatives. Methods: Secondary cases of PD were assessed both through a single informant (family history method) and through the study of each relative (family study method). The family study method was considered as the standard for comparison, and the sensitivity and specificity of the family history method were studied. Results: A total of 133 population-based case subjects and their 655 relatives were recruited, and 119 population-based control subjects and their 511 relatives. Sensitivity was 68% (95% CI = 47 to 85) for cases and 45% (95% CI = 17 to 77) for controls. Specificity was 99% (95% CI = 98 to 99) for cases and 100% (95% CI = 99 to 100) for controls. The odds ratio (OR) for family history of PD was 4.34 (95% CI = 1.63 to 11.58, p = 0.003) using the family history method and 1.86 (95% CI = 0.78 to 4.44, p = 0.16) using the family study method. The former significant OR more than doubled the latter not significant OR (relative bias = 133%). Bias was more pronounced for proxy interviews and for women informants, and when the relatives were siblings, were living, and were examined or had medical record documentation. Conclusions: Case subjects with PD (or their proxies) are more aware of PD among their first-degree relatives than control subjects (or their proxies); however, they overreport PD in relatives who are not affected. This causes a substantial family information bias.

AB - Objective: To study the validity of information provided by case and control subjects (or their proxies) about PD among their first-degree relatives. Methods: Secondary cases of PD were assessed both through a single informant (family history method) and through the study of each relative (family study method). The family study method was considered as the standard for comparison, and the sensitivity and specificity of the family history method were studied. Results: A total of 133 population-based case subjects and their 655 relatives were recruited, and 119 population-based control subjects and their 511 relatives. Sensitivity was 68% (95% CI = 47 to 85) for cases and 45% (95% CI = 17 to 77) for controls. Specificity was 99% (95% CI = 98 to 99) for cases and 100% (95% CI = 99 to 100) for controls. The odds ratio (OR) for family history of PD was 4.34 (95% CI = 1.63 to 11.58, p = 0.003) using the family history method and 1.86 (95% CI = 0.78 to 4.44, p = 0.16) using the family study method. The former significant OR more than doubled the latter not significant OR (relative bias = 133%). Bias was more pronounced for proxy interviews and for women informants, and when the relatives were siblings, were living, and were examined or had medical record documentation. Conclusions: Case subjects with PD (or their proxies) are more aware of PD among their first-degree relatives than control subjects (or their proxies); however, they overreport PD in relatives who are not affected. This causes a substantial family information bias.

UR - http://www.scopus.com/inward/record.url?scp=0038692023&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0038692023&partnerID=8YFLogxK

M3 - Article

C2 - 12847149

AN - SCOPUS:0038692023

VL - 61

SP - 11

EP - 17

JO - Neurology

JF - Neurology

SN - 0028-3878

IS - 1

ER -