BACKGROUND: In Alzheimer's disease (AD) clinical trials, including trials enrolling patients with mild cognitive impairment (MCI), participants must enroll with a study partner (SP). SPs ensure compliance and are a source of study data, including assessments of the participant's cognition and function. Consistency in SP reporting is essential to trial data integrity. METHODS: We quantified SP replacement and its impact on bias and variance of SP-reported AD Cooperative Study Activities of Daily Living for MCI (ADCS-ADL-MCI) in the ADCS Vitamin E/Donepezil MCI Trial. We used logistic regression to estimate the association between SP type (spouse or non-spouse) and the odds of experiencing SP change. We used generalized estimating equations to longitudinally model the differences in consecutively recorded ADCS-ADL-MCI scores as a function of whether SP change occurred. We used a similar model to quantify end-of-study change from baseline in ADCS-ADL-MCI scores. RESULTS: Among 768 participants, 40 (5%) experienced at least one SP change. We estimated that the odds of experiencing a SP change were 65% lower for spousal dyads when compared to non-spousal dyads (odds ratio [OR] = 0.35
95% confidence interval [CI]: [0.18-0.67]). Compared to those with a consistent SP, participants who experienced a SP change had, on average, a consecutive visit absolute score difference that was 1.60 points greater in magnitude (95% CI: [0.62-2.57]), suggesting greater volatility. ADCS-ADL-MCI scores were neither systematically higher nor lower when SP change occurred, on average (-0.23
95% CI: [-1.60, 1.14]), suggesting minimal bias. The estimated difference in variance for end-of-study change from baseline ADCS-ADL-MCI was observed to be higher for those with SP change compared to those without, but the difference was not statistically significant (1.29
95% CI: [0.47-1.17]). CONCLUSION: SP replacement occurred for a meaningful number of participants but did not result in systematic bias on a functional outcome in this trial, but it did increase variability.