Department of Communication Sciences and Disorders, The University of Iowa, Iowa City, Iowa.
Department of Biostatistics, The University of Iowa, Iowa City, Iowa.
J Am Acad Audiol. 2022 Mar;33(3):158-169. doi: 10.1055/a-1674-0060. Epub 2021 Oct 20.
Ecological momentary assessment (EMA) often places high physical and mental burden on research participants compared with retrospective self-reports. The high burden could result in noncompliance with the EMA sampling scheme protocol. It has been a concern that certain types of participants could be more likely to have low compliance, such as those who have severe hearing loss and poor speech recognition performance, are employed, are not familiar with technologies used to implement EMA (e.g., smartphones), and have poorer cognitive abilities. Noncompliance dependent on personal characteristics could negatively impact the generalizability of EMA research.
This article aims to determine personal characteristics associated with EMA compliance in a group of adult cochlear implant (CI) candidates and users.
An observational study.
Fifty-eight adults who were either scheduled to received CIs or were experienced CI users completed the study.
Participants conducted smartphone-based EMA designed to assess an individual's daily auditory ecology for 1 week. EMA compliance was quantified using two metrics: the number of completed surveys and the response rate to the notification delivered by the EMA app. Personal characteristics (i.e., predictors) included age, gender, CI status (candidate or user), employment status (employed or not employed), smartphone ownership, speech recognition performance, social network size, level of depressive symptoms, and neurocognitive abilities. A word recognition test, questionnaires, and a test battery of neurocognitive assessments were used to measure the predictors. We used negative binomial regression and logistic mixed models to determine the factors associated with the number of completed surveys and the response rate, respectively. We hypothesized that, for example, employed participants with poorer speech recognition performance would have lower compliance.
Contrary to the hypothesis, word recognition score was negatively associated with the number of completed surveys ( = 0.022). Holding all other variables constant, a 10-point (i.e., 10%) word recognition score decrease was associated with an 11% increase in the number of completed surveys. For the response rate, employment status was the only significant predictor ( < 0.0001). Consistent with our hypothesis, the odds of responding to EMA notifications for those who are not employed are 82% higher than the odds for those who are employed. No other studied personal characteristic was associated with compliance.
For CI candidates and users, EMA compliance could be affected by personal characteristics such as speech recognition performance and employment status. Because (1) participants with poorer speech recognition performance do not necessarily have lower compliance and (2) most personal characteristics investigated in the present study (e.g., age, gender, smartphone ownership, and neurocognitive abilities) do not predict compliance, a wide range of participants could successfully conduct smartphone-based EMA.
与回顾性自我报告相比，生态瞬时评估（EMA）通常会给研究参与者带来更高的身心负担。这种高负担可能导致参与者不遵守 EMA 抽样方案协议。人们一直担心某些类型的参与者更有可能不遵守规定，例如那些有严重听力损失和言语识别能力差、受雇、不熟悉用于实施 EMA 的技术（例如智能手机）以及认知能力较差的参与者。依赖个人特征的不遵守规定可能会对 EMA 研究的普遍性产生负面影响。
本文旨在确定一组成年人工耳蜗植入（CI）候选人和使用者中与 EMA 依从性相关的个人特征。
观察性研究。
58 名计划接受 CI 或经验丰富的 CI 用户的成年人完成了这项研究。
参与者使用基于智能手机的 EMA 进行了为期一周的个人日常听觉环境评估。使用两个指标来量化 EMA 依从性：完成的调查数量和 EMA 应用程序发送通知的响应率。个人特征（即预测因素）包括年龄、性别、CI 状态（候选者或使用者）、就业状况（受雇或未受雇）、智能手机拥有情况、言语识别能力、社交网络规模、抑郁症状水平和神经认知能力。使用单词识别测试、问卷和神经认知评估测试来测量预测因素。我们使用负二项回归和逻辑混合模型分别确定与完成调查数量和响应率相关的因素。我们假设，例如，言语识别能力较差的在职参与者的依从性较低。
与假设相反，单词识别分数与完成的调查数量呈负相关（ = 0.022）。在固定其他所有变量的情况下，单词识别分数降低 10 分（即 10%）与完成的调查数量增加 11%相关。对于响应率，就业状况是唯一显著的预测因素（ < 0.0001）。与我们的假设一致，未受雇者对 EMA 通知的响应概率比受雇者高 82%。没有其他研究的个人特征与依从性相关。
对于 CI 候选人和使用者，EMA 依从性可能会受到个人特征的影响，例如言语识别能力和就业状况。由于（1）言语识别能力较差的参与者不一定依从性较低，并且（2）本研究中调查的大多数个人特征（例如年龄、性别、智能手机拥有情况和神经认知能力）都不能预测依从性，因此可以成功地进行基于智能手机的 EMA 调查。
