Response to Comment on: “Nocturnal Hypoglycemia: Characterization with Continuous Glucose Monitoring in a Real-World Setting”

To the Editor,

We thank Nasralla et al. ¹ for their interest in our article ² and for highlighting the inherent challenges associated with analyzing real-world data (RWD) from mobile health applications. We appreciate the opportunity to clarify how these standard limitations were addressed in our study design.

Regarding the first point raised by the correspondents on selection bias, we agree that app users tend to be more technologically engaged and may demonstrate different self-care behaviors compared with the general population. We explicitly acknowledged this in our discussion, noting that users must have a certain level of digital affinity and smartphone access. However, we maintain that the value of this dataset lies in its scale and global reach, offering insights into glycemic patterns across diverse health care systems that randomized controlled trials cannot easily replicate.

Second, regarding the definition of the nocturnal period, the correspondents correctly note that fixed time windows (00:00–06:00) may misclassify hypoglycemia occurring during sleep outside these hours. While we agree that integrating wearable sleep-tracker data would offer superior precision—as demonstrated in the study by Martine-Edith et al. ³ cited by the correspondents—such data were not available for this retrospective cohort. In the absence of biometric sleep data, adhering to the standard consensus time window was necessary to ensure reproducibility and comparability with existing literature.

Last, regarding unadjusted covariates (e.g., meal timing, alcohol intake, sleep quality), RWD analysis is necessarily limited to the variables consistently captured in routine practice. While adjusting for these granular behavioral factors would be ideal, they are frequently sparsely recorded in real-world settings. We prioritized a robust sample size over a smaller, highly filtered cohort with complete behavioral logs, as this aligns with our objective to assess outcomes in a broad, naturalistic setting.

We view these points as inherent trade-offs characteristic of RWD research. These findings are intended to complement, not replace, the controlled precision of clinical trials.