Sponsors who want to use wearable devices in clinical trials should pay special attention to issues such as ensuring the clinical trials are blinded, data integrity, and whether the devices are fit-for-purpose, according to Guangxing Ken Wang, a statistician at the US Food and Drug Administration’s (FDA) Center for Devices and Radiological Health. He also emphasized that sponsors should consider verifying and validating the devices, and prespecifying their trial endpoints.
Wang spoke at the FDA/AdvaMed Medical Device Statistical Issues Conference on 2 April, where he said that digital health technologies (DHTs) such as accelerometers, continuous glucose monitors (CGMs), and inertial measurements units (IMUs) are increasingly being used in clinical trials due to the fact researchers can use them to track patients remotely. While the use of accelerometers and CGMs greatly outpaced other DHTs, he noted IMUs in trials have increased five-fold in recent years based on data from clinicaltrials.gov.
Wang said that wearables tend to be used in two ways in clinical trials. In certain studies, he said they are the devices being evaluated and may need to conform to certain special controls. In other studies, he noted they are used to collect data to support approval of another device or drug.
In the latter scenario, Wang said FDA has published recommendations, such as its Digital Health Technologies for Remote Data Acquisition in Clinical Investigations Guidance, that sponsors should read when using wearables in clinical trials. When considering the design and operation of the wearable device, he said researchers should consider confounding factors, such as how interaction with patients could affect the study outcome.
Since wearable devices may allow for two-way communication where researchers may be able to tell users to change their behavior based on the data they receive, Wang said they should ensure that the data is blinded and masked to ensure data integrity. He also noted that wearable devices have increased chances of failing because they continuously monitor and share data, which sponsors should account for.
“The longer the device tracks the patient, the more opportunity for the device to fail,” said Wang. “In that case, missing data is an important issue for wearable device data.”
He said sponsors may want to add features such as alerts that the user should change batteries and that the devices can handle large amounts of data to address concerns about potential missing data when designing them.
“If wearable devices are used in the study, it is important to verify and validate the wearable devices to make sure that the device used is fit-for-purpose and that the measures from the wearable devices are measured accurately and precisely,” said Wang. “We have to make sure that the clinical event or characteristic to be assessed is consistently and properly measured in the population of interest.”
When verifying and validating wearables used in clinical studies, Wang noted it’s important to conduct a study comparing the device to the predicate device if that is an option and evaluate factors such as the device placement and physical interferences that may affect the measurements taken by the device. He also added that sponsors should evaluate the calibration process, ensure the measurements taken by the device are consistent across devices from multiple brands’ and evaluate and justify potential differences in data collected remotely and in the clinic.
When considering endpoints derived from wearable devices, Wang said sponsors should follow the same principles for developing endpoints captured by other means meaning that they should be clinically meaningful and adequate data is collected, and the endpoint definition should detail the type of assessment being made, the timing of those assessments, and the tools used to measure the endpoints.
“Even though wearable devices can provide complex data and the opportunity to derive novel endpoints, it is still possible to use wearable devices to establish endpoints, for example, blood pressure or glucose value,” said Wang. “In that situation, the established endpoints from those wearable device data may not need new justification, but the verification and validation of the device is still needed.”
“Because of the complex nature of these wearable device data, it is possible to derive novel endpoints that may provide additional insight into participant functional performance,” he added.
Wang also talked about the statistical analysis and trial design considerations sponsors should keep in mind when conducting clinical trials that use wearable devices. More specifically, he said that all the study arms should use the same data collection method, the endpoints and their data sources should be prespecified in the statistical analysis plan, and the design should try to minimize potential missing data and data quality issues.
To continue reading this article please go to RAPS .