Is it safe to ask people about suicidal thoughts several times a day? Researchers and ethics committees have long raised concerns that repeated questions might worsen distress or suicidal ideation. Yet suicidal thoughts can change within hours, and occasional interviews or weekly questionnaires may miss important short-term fluctuations.
The SAFE study examined both questions. Adults with current suicidal ideation reported on their mood, thoughts, and suicidal ideation four times a day for three weeks using the Avicenna app (then Ethica). Most participants then continued with weekly assessments for a year. Alongside studying suicidal ideation itself, the researchers examined whether repeated assessments were feasible and acceptable and whether participants reported negative effects.
The study was conducted by the Department of Clinical Psychology at Leiden University in the Netherlands. Participants were recruited between August 2020 and September 2022.
Frequent assessments can reveal short-term changes in suicidal ideation, but researchers need a clear safety protocol and a plan for supporting participants who find the process difficult.
Study at a Glance
| Study feature | Detail |
|---|---|
| Population | Adults with a suicide attempt and/or active suicidal ideation in the past year |
| Recruitment | Community flyers and social media, plus mental health treatment centers in the Leiden and The Hague area |
| Enrolled | 82 participants; 81 completed the three-week EMA period |
| Data collection | August 2020 to September 2022 |
| EMA | Four scheduled assessments daily for 20 days, plus a final morning assessment on day 21 (81 scheduled assessments) |
| Response rate | Mean 78% of scheduled assessments; median 84% |
| Sleep | Sleep questions in the morning survey, plus around-the-clock actigraphy with a wrist-worn actigraph (CamNtech MotionWatch 8), collected separately from Avicenna |
| Follow-up | 72 participants (89%) continued into a 12-month period of weekly assessments |
| Safety | A personalized safety plan for every participant; the general practitioner and/or treating specialist was informed of their involvement |
| Ethics | Medical Ethics Committee Leiden, Den Haag, Delft (NL71510.058.19) |
| Funding | Netherlands Organisation for Scientific Research (NWO) Research Talent Grant 406.18.521 |
How the Study Worked
Intake and a Safety Plan
People interested in the study could check their eligibility online, then attend an intake interview in person or online. The interview confirmed eligibility using structured clinical interviews and collected participants’ medical and psychiatric histories.
Before receiving study instructions, each participant worked with a researcher to develop a personalized safety plan. It listed coping strategies and support resources they could use during a suicidal crisis, including the Dutch suicide prevention line 113. Participants were told clearly that nobody would read their answers in real time. They were instructed to contact their general practitioner, treating specialist, or one of the listed services if they experienced a crisis. The team also sent a standard letter to each participant’s general practitioner and/or treating specialist.
At the end of the intake, participants were asked to install the Avicenna app and enable notifications. A researcher walked them through a demonstration survey so they could familiarize themselves with the process.
Four Surveys a Day for Three Weeks
The app prompted participants once in each of four time windows: 7 to 9 a.m., 12 to 2 p.m., 4 to 6 p.m., and 8 to 10 p.m. Prompts were delivered at random times within each window. If a survey was not opened, a reminder followed after 30 minutes. Participants had 180 minutes to complete the morning survey and 120 minutes to complete the other surveys.
Participants received four scheduled assessments per day during the first 20 days, followed by a final morning assessment on day 21, for 81 scheduled assessments in total.
Each survey covered the same core topics:
- Context: where the participant was, who they were with, and what they were doing
- Affect: happiness, calmness, sadness, anxiety, anger, guilt, and shame
- Thoughts: hopelessness, loneliness, burdensomeness, and optimism
- Suicidal ideation: desire to live, desire to die, suicidal thoughts, and intention to act on them
- Events and coping: stressful or positive events, coping strategies, and substance use
The morning survey added questions about the previous night’s sleep, and the evening survey asked about naps. Participants could also start an extra survey whenever they wanted, for example, to record a particularly good or bad moment.
On average, participants took about 2 minutes and 46 seconds to complete a survey.
Tracking Completion, Not Answers
The researchers did not review participants’ survey responses in real time. Instead, they monitored survey completion and contacted participants by phone if they had not completed any surveys for 72 hours.
These calls were primarily intended to troubleshoot technical problems. The researchers also had procedures for directing participants to their general practitioner, treating specialist, or crisis services if needed. The study report states that no call required this crisis-related intervention.
This approach was part of the study’s specific safety protocol. Participants knew in advance that their answers would not be monitored in real time and were given clear instructions for seeking help during a crisis.
A Year of Weekly Check-Ins
About a week after the intensive assessment period ended, each participant met with a researcher to review a personalized feedback report based on their own data. Participants had been informed about this report from the beginning of the study.
Seventy-two participants (89%) continued into a 12-month follow-up period. Every Sunday, the app released a short questionnaire that remained open for 48 hours, with reminders after 12, 24, and 36 hours. The questionnaire included questions about events during the previous week, including whether the participant had made a suicide attempt.
Did the Questions Make Things Worse?
This was the central question of the first SAFE paper.
The researchers found no systematic worsening in momentary ratings. Participants’ momentary ratings of positive affect, negative affect, and suicidal ideation did not increase or decrease systematically over the 21-day assessment period. A standard questionnaire also showed lower suicidal ideation after the EMA period than before it. However, fewer participants completed both assessments, so this result does not establish that repeated surveys reduced suicidal ideation.
Most participants rated the experience positively. Among the 58 participants who completed the feedback survey, 69% described the experience as positive overall, and 48% described it as “insightful.” One participant wrote: “A lot of work, but also provided insights and sometimes it gave comfort.”
Some participants reported negative effects. In the feedback survey, 22% said their mood had sometimes worsened after completing an EMA entry, 18% said the assessments had triggered suicidal thoughts, and 10% said their suicidal thoughts had worsened. These retrospective reports did not correspond to a systematic increase in momentary ratings across the assessment period.
The reported effects were associated with different participant characteristics. Participants with more borderline personality traits or posttraumatic stress disorder (PTSD) were more likely to report that the assessments had triggered suicidal ideation. Reports of worsening suicidal ideation were associated with higher baseline suicidal ideation, depression and anxiety symptoms, and more borderline personality traits.
All participants who completed the feedback survey, including those who reported negative effects, continued into the second phase of the study.
The authors emphasized that researchers should explain to participants that repeated assessments may have positive or negative effects. They also recommended planning appropriate support for participants who may find the questions difficult.
These findings support the feasibility and acceptability of intensive EMA under the study’s specific conditions. They do not establish that repeated assessments are safe for every participant or that a similar protocol can be used without appropriate safeguards.
One Dataset, Several Questions About Suicidal Ideation
The SAFE dataset has also been used to investigate how suicidal ideation relates to other experiences and how it changes over time.
- Examining contemporaneous and temporal associations of real-time suicidal ideation using network analysis (2024): Which thoughts and feelings occur alongside suicidal ideation, and which predict it at the next assessment? Passive suicidal ideation was associated with sadness, hopelessness, loneliness, and burdensomeness, while active suicidal ideation was associated with sadness and shame. Hopelessness predicted passive suicidal ideation at the next assessment, and passive suicidal ideation, in turn, predicted active suicidal ideation. These temporal associations do not establish causation.
- Sleep, hopelessness, and suicidal ideation (2024): Does a poor night’s sleep predict suicidal thoughts the next day? Poorer self-reported sleep and more fragmented sleep measured by actigraphy were associated with higher suicidal ideation the following day. Hopelessness the next morning explained part of the association between self-reported sleep and suicidal ideation.
- Digital phenotypes of real-time suicidal ideation (2025): Do people differ in how their suicidal thoughts fluctuate? The analysis identified three profiles that differed in the frequency, intensity, and variability of suicidal ideation. Using data from the year of weekly follow-ups, the researchers found that the two profiles characterized by more intense or more variable ideation were associated with higher odds of a suicide attempt during follow-up. The authors noted that the sample was small and that the clinical usefulness of these profiles still needs to be established.
Together, these papers illustrate how intensive longitudinal data can help researchers examine not only whether suicidal ideation occurs, but also how it relates to sleep, emotions, other thoughts, and subsequent outcomes.
In Their Words
We asked Liia Kivelä, the first author of the SAFE papers, about running the study.
What should a researcher planning a similar study know before getting started?
“My main advice would be to pilot your study thoroughly, so that you can better understand the participants’ perspectives. Small design choices can have a surprisingly big impact on participants’ experiences. For example, even rating scales that are psychometrically sound may not always feel intuitive to participants. One of the great opportunities of EMA is that we can study people’s experiences as they naturally unfold, so it’s important that the app and the questions also reflect how people are used to rating and reporting their experiences in everyday life. Otherwise, we’re missing a part of what makes EMA so valuable.”
Was anything about running the study easier or harder than expected?
“From a technical perspective, setting up the surveys and running the study was very straightforward.”
What did Avicenna make possible that would have been difficult otherwise?
“Suicidal thoughts can differ considerably between people, but they can also change for the same person, even within a single day. Being able to survey participants about their experiences several times a day allowed us to capture these fluctuations as they naturally happen in people’s daily lives, and better understand how our daily behaviour and experiences relate to how we feel.”
Research Team
Niki Antypa
Principal investigator and corresponding author
Department of Clinical Psychology, Institute of Psychology, Leiden University
Liia M. M. Kivelä
Researcher and first author of the SAFE papers
Department of Clinical Psychology, Institute of Psychology, Leiden University
Willem van der Does
Supervisor and co-author of the SAFE papers
Department of Clinical Psychology, Institute of Psychology, Leiden University
Leiden University Treatment and Expertise Centre (LUBEC)
Publications
- Examination of Acceptability, Feasibility, and Iatrogenic Effects of Ecological Momentary Assessment (EMA) of Suicidal Ideation. Kivelä, L. M. M., Fiß, F., van der Does, W., & Antypa, N. (2024). Assessment, 31(6), 1292–1308.
- Examining contemporaneous and temporal associations of real-time suicidal ideation using network analysis. Kivelä, L. M. M., Fried, E. I., van der Does, W., & Antypa, N. (2024). Psychological Medicine, 54(12), 3357–3365.
- Sleep, hopelessness, and suicidal ideation: An ecological momentary assessment and actigraphy study. Kivelä, L. M. M., van der Does, W., & Antypa, N. (2024). Journal of Psychiatric Research, 177, 46–52.
- Digital phenotypes of real-time suicidal ideation: Correlates and consequences. Kivelä, L. M. M., van der Does, A. J. W., Gilissen, R., & Antypa, N. (2025). Acta Psychiatrica Scandinavica, 151(3), 375–387.
Planning a Similar Study?
Frequent assessments can help researchers capture changes in experiences and symptoms that occasional surveys may miss. But in sensitive research, the study protocol matters as much as the assessment schedule. SAFE paired intensive data collection with personalized safety plans, clear instructions for seeking help, and a process for following up when participants stopped responding.
Avicenna supports many of the data collection elements used in intensive longitudinal studies. Explore the documentation for Surveys, Time Triggering Logic, Expiry Time, and Notifications to learn how to schedule assessments, set completion windows, and send reminders. User Triggering Logic can also let participants initiate additional surveys, while Activity Sessions helps researchers review survey completion and identify gaps.
Explore Avicenna’s features and integrations or talk to our team to see how you can build your study protocol.