In many clinical studies, patients may be asked to report their medication adherence, presence of side effects, substance use, and hospitalization information during the study period. However, the exact occurrence time of these recurrent events may not be available due to privacy protection, recall difficulty, or incomplete medical records. Instead, the only available information is whether the events of interest have occurred during the past period. In this paper, we call these incomplete recurrent events as repeated current status data. Currently, there are no valid standard methods for this kind of data. We propose to use the Andersen-Gill proportional intensity assumption to analyze such data. Specifically, we propose a maximum sieve likelihood approach for inference and we show that the proposed estimators for regression coefficients are consistent, asymptotically normal and attain semiparametric efficiency bounds. Simulation studies show that the proposed approach performs well with small sample sizes. Finally, our method is applied to study medication adherence in a clinical trial on non-psychotic major depressive disorder.

中文翻译：

---

重复当前状态数据的半参数回归分析 [

在许多临床研究中，患者可能会被要求报告他们在研究期间的服药依从性、副作用的存在、物质使用和住院信息。但是，由于隐私保护、召回困难或医疗记录不完整，这些反复发生的事件的确切发生时间可能无法获得。相反，唯一可用的信息是感兴趣的事件是否在过去期间发生过。在本文中，我们将这些不完整的重复事件称为重复的当前状态数据。目前，对于此类数据没有有效的标准方法。我们建议使用 Andersen-Gill 比例强度假设来分析此类数据。具体来说，我们提出了一种用于推理的最大筛似然方法，并且我们表明所提出的回归系数估计量是一致的、渐近正态的并达到半参数效率界限。模拟研究表明，所提出的方法在小样本量下表现良好。最后，我们的方法用于研究非精神病性重度抑郁症临床试验中的药物依从性。