Nishida Mikaela K, Nuño Michelle M, Grill Joshua D, Gillen Daniel L
Department of Statistics, Donald Bren School of Information and Computer Sciences University of California Irvine California USA.
Department of Population and Public Health Sciences University of Southern California Los Angeles California USA.
Alzheimers Dement (N Y). 2024 Nov 30;10(4):e70009. doi: 10.1002/trc2.70009. eCollection 2024 Oct-Dec.
In Alzheimer's disease (AD) clinical trials, participants must enroll with a study partner informant who completes validated study instruments. We hypothesized that mid-trial informant replacement impacts study data in industry-sponsored trials.
We conducted a retrospective analysis of two industry-sponsored AD clinical trials testing semagacestat in mild-to-moderate AD dementia. We assessed the relationships between informant replacement and Alzheimer's Disease Cooperative Study Activities of Daily Living (ADCS-ADL) scores. Using generalized estimating equations, we assessed bias and variability using mean (bias) and mean absolute (variance) change in ADCS-ADL between successive visits as outcomes. Both models adjusted for a priori-specified potential confounding variables including participant sex, age, informant type, trial, time, previous ADCS-ADL score, and region. To analyze the impact on end-of-study change-from-baseline results, we used an analysis of covariance model to estimate the association between replacement and end-of-study change-from-baseline in ADCS-ADL, in which we adjusted for participant sex, age, informant type, trial, baseline measurement, and region. We conducted an -test to compare the variances of this change.
Among = 2637 randomized participants, 69 participants (2.6%) experienced 78 occurrences of replacement. For visits standardized to be 3 months apart, the difference in mean between-visit change in ADCS-ADL was approximately -1.61 points (95% confidence interval [CI]: -3.79, 0.57; = 0.147), comparing participants who experienced replacement to similar participants who had stable informants. The difference in the mean between-visit absolute change was approximately 2.02 points (95% CI: 0.34, 3.70; = 0.019). We did not estimate a statistically significant difference in end-of-study change-from-baseline (Est. = -0.70 points; 95% CI: -5.88, 4.48; = 0.790) or a significant ratio of variances (Est. = 1.13; 95% CI: 0.67, 2.28; = 0.600) for participants with replacement compared to those with stable informants.
Informant replacement was associated with increased between-visit variability but had limited impact on overall trial outcomes.
Informant replacement occurred in 2.6% of participants in these industry trials.Informant replacement was associated with increased variance in acute Alzheimer's Disease Cooperative Study Activities of Daily Living reporting.Informant replacement had a limited impact on overall change-from-baseline outcomes.
在阿尔茨海默病（AD）临床试验中，参与者必须与完成经过验证的研究工具的研究伙伴信息提供者一起登记入组。我们假设试验中期信息提供者的更换会影响行业赞助试验中的研究数据。
我们对两项行业赞助的AD临床试验进行了回顾性分析，这两项试验在轻度至中度AD痴呆中测试了司美加群。我们评估了信息提供者更换与阿尔茨海默病协作研究日常生活活动（ADCS-ADL）评分之间的关系。使用广义估计方程，我们以连续访视之间ADCS-ADL的均值（偏差）和平均绝对（方差）变化为结果评估偏差和变异性。两个模型都对预先指定的潜在混杂变量进行了调整，包括参与者性别、年龄、信息提供者类型、试验、时间、先前的ADCS-ADL评分和地区。为了分析对研究结束时基线变化结果的影响，我们使用协方差分析模型来估计更换与研究结束时ADCS-ADL基线变化之间的关联，其中我们对参与者性别、年龄、信息提供者类型、试验、基线测量和地区进行了调整。我们进行了F检验以比较这种变化的方差。
在2637名随机分组的参与者中，69名参与者（2.6%）经历了78次更换。对于标准化为相隔3个月的访视，与信息提供者稳定的类似参与者相比，经历更换的参与者在ADCS-ADL访视间平均变化的差异约为-1.61分（95%置信区间[CI]：-3.79，0.57；P = 0.147）。访视间平均绝对变化的差异约为2.02分（95%CI：0.34，3.70；P = 0.019）。我们没有估计出与信息提供者稳定的参与者相比，经历更换的参与者在研究结束时基线变化有统计学显著差异（估计值=-0.70分；95%CI：-5.88，4.48；P = 0.790）或方差的显著比率（估计值=1.13；95%CI：0.67，2.28；P = 0.600）。
信息提供者更换与访视间变异性增加相关，但对总体试验结果影响有限。
在这些行业试验中，2.6%的参与者发生了信息提供者更换。信息提供者更换与急性阿尔茨海默病协作研究日常生活活动报告中的方差增加相关。信息提供者更换对总体基线变化结果影响有限。
