b'CHAPTER 2 | THE AGING BRAINProxy Informants questionnaire about the mainscores from the HRS. The cognitively normal Bias arises in surveys whensurvey participant. Thegroup was further stratified by age (70 to 79, and respondents decline orinformant is most often a80 or older) and sex in order to ensure adequate are unable to participatefamily member involved innumbers in each of these subgroups. because they are toocaregiving. The informantA nurse and a neuropsychology technician sick. This can resultalso serves as the proxyconducted interviews and assessments in the in underestimatingrespondent if one washomes of ADAMS participants. The information the true prevalence ofneeded. collected in the home was then reviewed by a health conditions likePotential racial biasboard of qualified scientists and medical doctors dementia. To addressin cognitive measures is ato determine a diagnosis. The medical team deter-this form of attrition bias,concern both for self-reportedmined three general categories: normal cognition, family members of partici- and proxy information. Potter etCIND and dementia. Dementia often arises from pants in the HRS and ADAMS areal. (2009) investigate this in ADAMSdifferent medical conditions, so dementia was often called upon to provide surveyand find no evidence of racial bias infurther categorized as dementia, AD and vascular information on behalf of participants, servingself-reported information. They also compare in- dementia (VaD).as proxy respondents. Approximately 10% offormant reports of African Americans and WhitesUsing these landmark data, Plassman et al. HRS interviews are obtained through proxies,in ADAMS. Results for dementia reveal no signif- (2007) report the first national estimates of the and a large proportion of ADAMS respondentsicant racial differences. However, African Ameri- prevalence of dementia and dementia subtypes in with dementia provide information throughcan informants are much less likely than Whitesindividuals over the age of 70. proxies. Weir et al. (2011) compare results fromto identify symptoms corresponding to CIND in HRS to those from the English Longitudinaltheir family members with this condition.Survey of Ageing (ELSA), which is based on the HRS but uses proxies much less often. InThe First National Data the HRS, proxy respondents are asked a seriesReliable data on dementia incidence and of questions about the respondents change inprevalence is critical for establishing memory for various types of information in thebenchmarks against which to measure last two years. Use of proxies in the HRS leadspopulation-wide changes over time. ADAMS to significantly higher overall response ratesprovides the first national data on the and virtually eliminates attrition bias in theprevalence of clinically diagnosed dementia estimation of cognitive function.and CIND. A random sample of 1,770 HRS participants aged 70 years or older in 2001 Informant Reports were invited to take part in ADAMS. Another source of information that can be usedThe sample was stratified on five to improve the accuracy of dementia determi- levels of cognitive func-nation is the informant report. ADAMS asks ationing based close family member to complete an additionalon cognitive 39'