Dropouts and refusals in observational studies

Abstract

The success of prevention trials of Alzheimer disease and other dementias (AD/dementia) hinges on their ability to recruit and retain at-risk study populations. Losing subjects is a threat to the power to detect a treatment effect and, potentially, to the validity of these studies. Observational cohort studies accumulate data around participant outcomes that can help to inform the design of future prevention trials. Our objectives were to investigate the rates of refusal and dropout within observational cohort studies and to evaluate their characteristics and impact. This study examined data from the Canadian Cohort Study of Cognitive Impairment and Related Dementias (ACCORD), a 2-year observational cohort study of patients newly referred to dementia clinics. The sample included 124 Not Cognitively Impaired (NCI) and 342 Cognitively Impaired Not Demented (CIND) subjects. Subjects who refused initial neuropsychological (NP) testing were compared to those who completed NP testing and subjects who dropped out to those who attended follow-up. Refusal was common, with 40% of subjects not completing neuropsychological testing at baseline. Dropout was also substantial, with 55% lost to the 2-year follow-up. Subjects who refused NP testing were significantly older and less educated. CIND refusers had lower cognitive and functional scores at entry and a 2-year progression rate to dementia twice as high as that of non-refusers. CIND dropouts also had lower baseline cognitive and functional scores. These observations suggest that dropouts and refusals in prevention trials include those subjects who are at high risk for progression to AD/dementia. Targeted strategies to retain these individuals within prevention studies will be needed to achieve sufficient study power and validity.