Menu Back to Poster-Presentations-Details

W-25: Bring your own Wearable (BYOW): Considerations for Clinical Research

Poster Presenter

Marie McCarthy

Digital Endpoint Lead
Novartis
Ireland

Objectives

This poster quantifies the penetration of fitness trackers in clinical trials, identifies considerations for a future BYOW model and proposes lessons learned from the principles of measurement equivalence for patient-reported outcomes using BYOD should be considered before exploring this approach.

Method

1) The scientific approach used to validate the BYOD model was used as a framework to determine criteria for assessing the suitability of BYOW. 2) Using generic term “Fitness tracker” and common brand names; clinicatrials.gov was mined to quantify adoption of these devices in clinical studies.

Results

1) Several key components of assessing equivalence in the BYOD literature was found to have relevance for considering BYOW: (i) The need to maintain the important measurement properties of the outcome measure across the range of devices, and (ii) identification of parameters that influence the measurement properties across devices. However, BYOW also poses unique questions such as the suitability of processed or raw data to determine device equivalence, and the ability to accurately compare data between studies using different devices. 2) 534 trials were identified using the search criteria. Of these studies, 83% were initiated in the last 5 years (2014-2018). The majority of the studies were behavioural studies (279, 52%), and involved some form of intervention (476, 89%). The search revealed 18 pharmaceutical clinical trials across all phases, of which the data generated from the fitness trackers was used to support both primary outcomes (2 of 18) and secondary outcomes (13 of 18).

Conclusion

: It is clear from the research that in the last 5 years there has been a small but sizable use of fitness trackers in clinical research. These trackers are being used for a number of different purposes, including supporting behavioural changes and the generation of data to support regulatory decision making. Part of the rationale for the use of fitness trackers is widespread personal adoption of these devices. In addition, a number of device manufacturers are pursuing medical device labelling claims as evidenced by the pre-certification program recently established by the FDA. This opens up the possibility of these devices increasingly being used in clinical research and, perhaps, the potential of subjects using their own device to generate data. As with BYOD, it is likely that patients may want to use their own wearables in research and this BYOW model may be a cost-effective patient-centric alternative. However, before this approach can be adopted there are a number of considerations and questions posed: (i) Is there sufficient evidence that the data generated is sufficiently equivalent across devices? (ii) Does this apply to processed data provided by devices (e.g., step counts) or only to raw data (e.g., accelerations)? (iii) Can we tolerate differences between device models/brands when considering within-subject changes as opposed to absolute values, and is it possible to define meaningful change in this scenario? (iv) If multiple studies or different programmes collect the same endpoint but use different devices – are the results easily comparable? These considerations are important as we consider the potential of BYOW in clinical research.