Skip to content

Consensus Statement: Conducting Digital Monitoring Studies of People at Suicide Risk

Key findings

  • A recent consensus meeting considered the safe and ethical conduct of digital monitoring studies (e.g., using smartphones or wearable biosensors) of people at risk of suicide and related behaviors
  • The 24 participants in the modified Delphi consensus-building process included leading scientists, clinicians, ethicists, legal experts, those with lived experience and other interested stakeholders
  • This summary reports 13 topics on which there was strong agreement, including that participants should not be excluded or removed from studies solely because of the high level of suicide risk
  • The consensus statement is designed to provide guidance for researchers, funding agencies and institutional review boards

Real-time digital monitoring studies of people at risk of suicide have recently increased exponentially in number. However, incorporating technologies such as smartphones and wearable biosensors presents unprecedented methodological, clinical, ethical and legal concerns.

Matthew K. Nock, PhD, a research scientist in the Department of Psychiatry at Massachusetts General Hospital, and colleagues recently held a consensus meeting about best practices for conducting these studies. Their statement in Psychiatric Research & Clinical Practice provides guidance for researchers, funding agencies and institutional review boards.

Modified Delphi Process

Twenty-four experts provided input into an anonymous survey, an in-person meeting with presentations and discussion, and a final survey. The expertise represented included science, clinical practice (inpatient/outpatient and child/adolescent/adult), bioethics, law, institutional review boards, computer science, statistics, funding agencies, and people with personal or lived experience with suicidal thoughts and behaviors.

This summary presents the points of strong agreement (≥80%).

Before Data Collection

  • Eligibility—Do not exclude participants solely because of a high level of clinical severity or suicide risk
  • Conditions for enrollment—There should be no conditions that participants must accept in order to enroll or remain in the study, such as agreeing to be in treatment or calling a hotline when at high risk
  • Informed consent—Give participants explicit information about certain key elements of study procedures (enumerated in the article). There was no strong agreement that participants should be informed about what specific monitoring and interventions may occur
  • Contact information—Collect and retain (during the monitoring period) contact information from the participant and at least one collateral person so participants can be reached during periods of perceived elevated risk
  • Technological considerations—Determine what to do if technology fails; test the alert system; in a standardized manner, review items of incidental data collection (medical information, voice, video) being obtained and the process of alerting and collecting data; and decide how to store and potentially share location data
  • Safety procedures—Provide participants with emergency contact information, determine what will trigger further risk assessment or intervention, and train study staff in assessing and responding to suicidal ideation and behavior

During Data Collection

  • Frequency of review—Review participant survey responses at least once every weekday
  • Determining risk level—Collect data about the current level of desire to die, level of intent to die, presence of a suicide plan and access to the planned method
  • Response window—Respond to participants determined to be at imminent risk of suicide within 24 hours of learning of the risk (72% of the panel indicated such a response should be made within 12 hours)
  • Type of intervention—In situations of high risk, reach out to the participant directly to conduct a risk assessment. Before monitoring begins, formulate a personalized risk plan with each participant to guide procedures for intervening if the participant does not respond to initial calls, texts or emails from the team. In studies involving youth, contact a parent in instances of non-response by a high-risk adolescent. Automated interventions are sufficient for low-risk and moderate-risk situations
  • Participant removal—No participant should be removed (temporarily or permanently) from the study by the research team (vs. an individual participant's wish to discontinue) due to elevated risk of suicide or clinical worsening
  • Data security—Store data in a de-identified form on secure web-based platforms and in compliance with HIPAA guidelines
  • Data safety and monitoring board (DSMB)—If a DSMB is used, it should include at least one person with expertise in managing suicide risk

Updates to the consensus guidance are expected as more data become available, and future meetings should reconsider issues on which no consensus could be reached, including what constitutes a high or imminent risk of suicide.

Learn more about the Department of Psychiatry

Learn more about research in the Department of Psychiatry

Related topics


By monitoring people via smartphone several times per day, Mass General researchers became the first to show that there are distinct patterns of suicidal thinking, which may have implications for suicide prevention efforts.


Using electronic health records to predict who needs suicide prevention interventions has been criticized as impractical. But decision modeling by Eric L. Ross, MD, and Jordan W. Smoller, MD, ScD, of the Department of Psychiatry, and colleagues suggests some existing models are cost-effective and warrant pilot studies.