Trial Review

The safety and scientific validity of this study is the responsibility of the study sponsor and investigators. Listing a study does not mean it has been endorsed by the ANZCTR. Before participating in a study, talk to your health care provider and refer to this information for consumers
Trial registered on ANZCTR


Registration number
ACTRN12621001248853
Ethics application status
Approved
Date submitted
2/08/2021
Date registered
15/09/2021
Date last updated
11/03/2022
Date data sharing statement initially provided
15/09/2021
Date results information initially provided
14/01/2022
Type of registration
Prospectively registered

Titles & IDs
Public title
The MyMood&Me Project: Using text data to infer mental health
Scientific title
The MyMood&Me Project: Assessing the feasibility of text data to identify linguistic markers of mental health in adults with depression and anxiety
Secondary ID [1] 304905 0
None
Universal Trial Number (UTN)
U1111-1268-3002
Trial acronym
Linked study record

Health condition
Health condition(s) or problem(s) studied:
Depression 323042 0
Anxiety 323043 0
Condition category
Condition code
Mental Health 320611 320611 0 0
Anxiety
Mental Health 320612 320612 0 0
Depression

Intervention/exposure
Study type
Interventional
Description of intervention(s) / exposure
The MyMood&Me experimental pilot aims to explore the role of text data for producing reliable and valid linguistic markers of depression and anxiety among adults with mental illness. Over 9 weeks (i.e., 63 days total: 56 days of intervention, +7 days to complete final endpoint survey), participants will be randomly allocated to receive a random sequence of four repetitive stimulated writing tasks, and a set sequence of four one-off stimulated writing tasks. The total intensity and frequency of writing tasks will be the same for all participants; however, the order in which the repetitive tasks are completed will be different. There are four intervention periods in the current study, each lasting 14 days, with two types of writing tasks scheduled in each intervention period. All the scheduled writing tasks and surveys will be delivered online via a website on the Black Dog Institute Research Engine.

Participants will also be asked to complete five fortnightly mental health surveys on days 1, 14, 28, 42, 56 (i.e., at the beginning of each 14-day intervention period). The mental health survey includes standardised, self-reported psychometric scales for depression (Patient Health Questionnaire-9), and anxiety (Generalised Anxiety Disorder Scale -7). At baseline only, participants will complete demographics questionnaires, motivations for participation, and a short, validated screener for personality disorders. At baseline and final endpoint, participants will complete the rumination and mastery questionnaires. The survey will take no more than 10 minutes to complete. Each survey will stay open for 48 hours. Participants will receive an initial invitation (via email and SMS) to complete, followed by two reminders (sent via email and SMS).

The one-off writing tasks will be scheduled to be completed at the same time as the corresponding fortnightly mental health survey (i.e., on days 14, 28, 42 and 56). The writing task will be administered first to reduce any carryover effect of the symptom scales. There is no washout period as there is nil to limited evidence of a carryover effect for the proposed writing interventions. The one-off writing tasks used in the trial include 1) a personal writing task in which participants are instructed to write about their physical and personality traits, their hobbies, etc, (Task W) 2) a neutral writing task where participants will be instructed to write about how they spend time with their friends (Task X), 3) a narrative imagery task where participants will be instructed to write what they think is happening in the provided image (Task Y), and 4) a holidaying letter to a friend in which participants are asked imagine they're on holiday and writing a letter to their friend to convince them to come over (Task Z). Participants will be advised to allocate at least 5 minutes to each one-off writing task. The order of the one-off writing tasks will be set for all participants (i.e., Task WXYZ) and will stay open for 48 hours. As it forms part of the fortnightly mental health survey, participants will receive an initial invitation (via email and SMS) to complete, followed by two reminders (sent via email and SMS to non-completers). The total duration of the fortnightly mental health survey + the one-off writing task will take an estimated 15 minutes to complete, with participants able to take longer if they wish.

The repetitive stimulated writing tasks are scheduled to be delivered on a set sequence/schedule in the 14-day intervention period. Repetitive writing tasks used in this trial include 1) SMS retrieval (Task A), 2) a social media task (Task B), 3) an emotion diary (Task C), and 4) an anchored negative event task (task D). For the SMS retrieval task, participants will be asked to retrieve a recent private message, this request will be sent every 2 days for a total of 5 SMS requests. The social media task asks participants to write a hypothetical post on social media, this request will be sent every 3 days for a total of 4 written social media posts. The emotion diary task asks participants to write about an event that impacted their mood, this request will be sent every 4 days for a total of 3 diary entries. Lastly, the anchored negative task will ask participants to write about their deepest thoughts and feelings on an extremely important emotional issue that has affected them and their life. This task will be asked every 4 days for a total of 3 anchored negative event requests. The mood monitoring questionnaire will be delivered prior to, and immediately after, each emotion diary task and negatively anchored writing tasks to monitor any impacts on individuals’ mood. Participants will be advised to allocate 5 minutes to each of the scheduled writing tasks. Each task will remain open for 24 hours, with participants receiving 3 invitations to complete (initial invitation sent in the AM, two reminders sent to non-completers in the PM).

Adherence to the writing tasks will not be monitored throughout the study period. Participants will receive scheduled reminders (up to 3) to complete the writing tasks and surveys. An acceptability questionnaire will be delivered after participants complete each writing task for the first time (i.e, total of 8 acceptability questionnaires) to measure the level of acceptability for each task.
Intervention code [1] 321301 0
Treatment: Other
Comparator / control treatment
This is a pilot randomised crossover design. Participants act as their own control, with the reference comparator task being the neutral writing task.
Control group
Active

Outcomes
Primary outcome [1] 328424 0
Depressive Symptoms: This will be measured on 5 occasions using the 9-item self-report Patient Health Questionnaire-9 (PHQ-9; Kroenke, 2001). The PHQ assesses the presence of depressive symptoms in the past two weeks. Item scores are summed to produce a total score, which can be used to indicate ‘nil to mild’ symptoms (scores: 0-9), ‘moderate’ symptoms (scores: 10-14), and ‘moderately severe’ to ‘severe’ symptoms (scores: 15+). Suicidal ideation will be measured using the single item-9 from the PHQ-9, The survey will available to participants for 7 days from the start of each time point.
Timepoint [1] 328424 0
Timepoint 1: Baseline (Day 1-7)
Timepoint 2: Week 2 (Day 14-21)
Timepoint 3: Week 4 (Day 28-35)
Timepoint 4: Week 6 (Day 42-49)
Timepoint 5: Endpoint (Day 56-63)
Primary outcome [2] 328425 0
Anxiety Symptoms: This will be measured on 5 occasions using the 7-item self-report Generalised Anxiety Disorder Scale -7 (GAD-7; Spitzer, 2006). The GAD assesses the presence of anxiety symptoms in the past two weeks. Item scores are summed to produce a total score, which can be used to indicate ‘nil to mild’ symptoms (scores: 0-9), ‘moderate’ symptoms (scores: 10-14), and ‘moderately severe’ to ‘severe’ symptoms (scores: 15+). The survey will available to participants for 7 days from the start of each time point.
Timepoint [2] 328425 0
Timepoint 1: Baseline (Day 1-7)
Timepoint 2: Week 2 (Day 14-21)
Timepoint 3: Week 4 (Day 28-35)
Timepoint 4: Week 6 (Day 42-49)
Timepoint 5: Endpoint (Day 56-63)
Secondary outcome [1] 399012 0
Rumination: Rumination is a method of coping with negative mood that involves self-focused attention. It is characterised by self-reflection as well as a repetitive and passive focus on one’s negative emotions. In the current study, rumination is measured to understand the relationship between mental health, linguistic markers, and engagement in writing tasks. This study will measure rumination at baseline and endpoint using the short form Ruminative Response Scale (RRS; Treynor, W., Gonzalez, R., & Nolen-Hoeksema, S, 2003). The RRS-short form is a self-administered questionnaire of 10 items describing two dimensions of rumination: brooding and reflection. For each item, each subject indicates the frequency of each event on a 4-point scale ranging from almost never (1) to almost always (4). Scores may range between 10 and 40, with higher scores indicating higher levels of ruminative response styles. The survey will available to participants for 7 days from the start of each time point.
Timepoint [1] 399012 0
Timepoint 1: Baseline (Day 1-7)
Timepoint 5: Endpoint (Day 56-63)
Secondary outcome [2] 399020 0
Mastery: This will be measured at baseline and endpoint using the Perceived Mastery Scale (PM; Pearlin & Schooler, 1978) and will provide a sense of mastery versus helplessness that participants feel about their lives. Participants will be asked to rate how much they agree with 7 statements related to mastery, answered on a 6-point Likert scale ranging from (1) strongly disagree to (6) strongly agree, with a final option (neither agree or disagree). Items are summed, yielding a range from 7 to 28. Higher scores indicate greater levels of mastery. The survey will available to participants for 7 days from the start of each time point.
Timepoint [2] 399020 0
Timepoint 1: Baseline (Day 1-7)
Timepoint 5: Endpoint (Day 56-63)
Secondary outcome [3] 399029 0
Mood Monitoring: This will be measured using the 6-item short form of the Multidimensional Mood Questionnaire (MDMQ). The MDMQ is a validated mood monitoring scale answered using a 7-point Likert scale. This scale will be delivered immediately before and after the negatively anchored repetitive writing tasks. Participants are asked to indicate how they feel in this very moment, in relation to fatigue, content, agitation. Within subjects, the scale has demonstrated the ability to differentiate three synchronized mood states. The three measures are also highly sensitive to capture changes in mood states.
Timepoint [3] 399029 0
Asked before and immediately after the negative anchored writing tasks (i.e., the Emotion Diary and the Expressive Writing Task). Delivered 6 times throughout the study period.
Secondary outcome [4] 399030 0
Acceptability: Measured through a 3-item questionnaire developed for this study. To determine the acceptability of the writing tasks and the study design participants will be asked to complete 3 questions at the end of each writing task: How easy or difficult was this task? How interesting was this task? How willing would you be to complete this task again? These questions will be answered using a 7-point Likert scale ranging from Not at all (1) to Extremely (7). This questionnaire will be delivered after each writing task is completed for the first time. There is a total of 8 acceptability questionnaires scheduled throughout the entire study period.
Timepoint [4] 399030 0
Asked at the end of each writing task, when scheduled for the first time only. This questionnaire will only be asked once for each of the writing tasks, delivered 8 times in total throughout the study.

Eligibility
Key inclusion criteria
1) Are currently 18 years of age or older (confirmed by self-report)
2) Have an active mobile phone number and email address (for receipt of study invitations and reminders)
3) Have at least moderate symptoms of depression or anxiety (i.e., total score greater than or equal 10 on the PHQ-9 or GAD-7).
4) Living in Australia
Minimum age
18 Years
Maximum age
No limit
Sex
Both males and females
Can healthy volunteers participate?
No
Key exclusion criteria
1) Report a suicide attempt in the past six months (i.e., Have you made a suicide attempt in the past six months? Answered ‘Yes’ or ‘No’)
2) Report extreme and unmanageable emotional distress (i.e., “Are you currently experiencing extreme and unmanageable emotional distress?” Answered ‘Yes’ or ‘No’)

Study design
Purpose of the study
Diagnosis
Allocation to intervention
Randomised controlled trial
Procedure for enrolling a subject and allocating the treatment (allocation concealment procedures)
Allocation will occur using the Black Dog Institute Research Engine, which is research trial software that automatically allocates participants to their sequence of intervention. Participants are allocated after providing their consent to participate. Participants and the research operations team will remain blinded to participants’ allocation until they complete the study.
Methods used to generate the sequence in which subjects will be randomised (sequence generation)
A random-number generator will be used to generate the sequence order of interventions. This will be conducted by the trial statistician who will not be involved in the day-to-day operations of the trial. The Black Dog Research Engine will then be configured to randomly allocate participants to receive a set sequence of writing tasks, using computerised sequence generation.
Masking / blinding
Blinded (masking used)
Who is / are masked / blinded?
The people receiving the treatment/s
The people administering the treatment/s

Intervention assignment
Crossover
Other design features
Phase
Not Applicable
Type of endpoint/s
Efficacy
Statistical methods / analysis
Sample size: The target sample size was based on detecting moderate correlations (r=0.3-0.5) between the linguistic features and mental health outcomes for depression and anxiety (primary outcome). Given that the study design is intensive, a conservative estimate of 40% attrition has been used. As such, the recruitment target for the trial is N=140.

Data analysis: Participants’ text data will be exported from the online survey platform and analysed using Linguistic Inquiry and Word Count (LIWC) (Pennebaker et al., 2007), a robust and widely validated tool for extracting linguistic features from text. This software analyses text and calculates the percentage of words that reflect different emotions, thinking styles, social concerns, and parts of speech captured by the LIWC program dictionary (Tausczik & Pennebaker, 2010). This will result in a set of 68 linguistic features for each post which will then be correlated with mental health scores. The tool also calculates the total number of words within the posts that match the program dictionary, reported as “dictionary words”. LIWC scores will be averaged across each text task, and then combined for an overall LIWC score. This will result in dataset consisting of participants’ symptom scores matched with their averaged LIWC scores for the same period. Latent Dirichlet Allocation (LDA) will be used for topic analysis. A generative model that allows sets of observations in text to be explored. It posits that each document is a mixture of a small number of topics and that each word is attributable to one of the document’s topics. Affective Norms for English Words (ANEW) will be used for emotional sentiment. This allocates a valence and arousal score for each English word. The dimension of valence ranges from highly positive to highly negative and arousal ranges from calming to agitating.

For the inferential modelling, bayesian context models and machine learning methods will be employed to extract, for each participant, the linguistic features, emotional sentiment, and topics that highly correlate with their mental health. Bivariate analyses will be undertaken to investigate the correlation between the linguistic features and symptom scores between individuals. We will perform multivariate analysis between multiple linguistic features using partial-least squares (PLS) regression. Because our past research has shown that these group-level inferences do not always generalise to intra-individual changes in symptom scores over time, we will test the PLS regression model constructed on group-level data on repeated measures of single participants using a two-staged approach. We will first use the group-level model to predict the symptom scores at each time point at which linguistic features were extracted and symptoms were assessed and correlated the predicted and observed symptom scores across time points for each participant. We will then compare the correlation coefficients estimated for each participant at the group level. To do this, we will convert the correlation coefficients using Fisher's z transformation and compare the z-scores against zero using a one-sample t-test.

Recruitment
Recruitment status
Completed
Date of first participant enrolment
Anticipated
4/10/2021
Actual
20/10/2021
Date of last participant enrolment
Anticipated
2/08/2022
Actual
27/10/2021
Date of last data collection
Anticipated
4/10/2022
Actual
27/12/2021
Sample size
Target
140
Accrual to date
Final
247
Recruitment in Australia
Recruitment state(s)
ACT,NSW,NT,QLD,SA,TAS,WA,VIC

Funding & Sponsors
Funding source category [1] 309281 0
Government body
Name [1] 309281 0
National Health and Medical Research Council
Country [1] 309281 0
Australia
Primary sponsor type
University
Name
University of New South Wales (UNSW)
Address
University of New South Wales
Sydney NSW 2052
Country
Australia
Secondary sponsor category [1] 310250 0
None
Name [1] 310250 0
Address [1] 310250 0
Country [1] 310250 0

Ethics approval
Ethics application status
Approved
Ethics committee name [1] 309117 0
University of New South Wales Human Research Ethics Committee
Ethics committee address [1] 309117 0
UNSW Research Ethics & Compliance Support
The University of New South Wales
Sydney
NSW 2052
Ethics committee country [1] 309117 0
Australia
Date submitted for ethics approval [1] 309117 0
24/05/2021
Approval date [1] 309117 0
09/07/2021
Ethics approval number [1] 309117 0
HC210397

Summary
Brief summary
Linguistic markers generated from within individuals’ text data may present a new way to identify individuals who are experiencing depression and anxiety. The MyMood&Me study aims to explore the potential of text data for producing reliable linguistic markers of depression, anxiety, and suicidal ideation among adults with mental illness. Our aim is to learn how whether the way individuals construct their sentences (i.e., how they write, rather than what they write) can indicate feelings of depression, anxiety, and thoughts of suicide.

To investigate this, adults with depression will be invited to participate in a 9-week randomised crossover pilot trial in which they complete fortnightly symptom scales for depression and anxiety (PHQ-9, GAD-7), while being allocated to a random sequence of writing tasks. The writing tasks are varied in their emotional anchoring, and participants are instructed to write freely for approximately ten minutes. Upon completion of the study, individuals’ linguistic features will be extracted from the text data using the LIWC tool. Bivariate and multivariate analyses will be performed to investigate the correlations between the linguistic features and symptoms between subjects. Multivariate analyses will be used to explore the relationship between symptoms and linguistic expression within subjects throughout the study period. Descriptive analyses will examine the impact of writing tasks on participant burden. Together, these results will provide initial evidence to support the use of text tasks for generating linguistic markers of depression, anxiety, and suicidal thoughts, and determine the type of text task (e.g., SMS, social media posts, expressive writing) most useful and engaging for participants to undertake.

If the brief writing tasks are deemed acceptable and engaging, this may present a new way to monitor participants’ response to therapy, including tracking recovery and relapse.
Trial website
Trial related presentations / publications
Public notes

Contacts
Principal investigator
Name 113034 0
Dr Bridianne O'Dea
Address 113034 0
Black Dog Institute
Hospital Road
Prince of Wales Hospital
Randwick, NSW, 2031
Country 113034 0
Australia
Phone 113034 0
+61 2 9382 8509
Fax 113034 0
Email 113034 0
[email protected]
Contact person for public queries
Name 113035 0
Dr Bridianne O'Dea
Address 113035 0
Black Dog Institute
Hospital Road
Prince of Wales Hospital
Randwick, NSW, 2031
Country 113035 0
Australia
Phone 113035 0
+61 2 9382 8509
Fax 113035 0
Email 113035 0
[email protected]
Contact person for scientific queries
Name 113036 0
Dr Bridianne O'Dea
Address 113036 0
Black Dog Institute
Hospital Road
Prince of Wales Hospital
Randwick, NSW, 2031
Country 113036 0
Australia
Phone 113036 0
+61 2 9382 8509
Fax 113036 0
Email 113036 0
[email protected]

Data sharing statement
Will individual participant data (IPD) for this trial be available (including data dictionaries)?
Yes
What data in particular will be shared?
The participant-level data that will be shared is: mental health scores, LIWC scores, sentiment analyses and topic analyses. The data custodian will remove individual identifiers (email address, mobile phone number, IP addresses) from the data. Only participants who have consented to this data sharing will be shared.
When will data be available (start and end dates)?
The participant-level data will become available upon completion of the study and after the primary analyses have been conducted and published. We expect this to be available from 1st January 2023 and will remain accessible for 15 years until 1st January 2038.
Available to whom?
The data will only be accessible to other researchers for studies related to this general research topic. Where possible, the researcher will ask others who wish to access the data for a copy of their ethics approval to do so before the data is shared for secondary research purposes. The researcher will maintain a copy of other researchers' ethics approval for their records.
Available for what types of analyses?
Researchers will be able to access the data that includes analyses related to the field of linguistics and mental health and/or markers of mental illness.
How or where can data be obtained?
Dr Bridianne O’Dea is the data custodian for this project. Any researchers interested in accessing the data for secondary research purposes will be required to contact the data custodian and request access using a data access request form. Dr Bridianne O'Dea can be contacted via email: [email protected] or via phone +61423366563.


What supporting documents are/will be available?

Doc. No.TypeCitationLinkEmailOther DetailsAttachment
12706Ethical approval    382494-(Uploaded-28-07-2021-16-10-51)-Study-related document.Pdf
12707Study protocol    382494-(Uploaded-07-03-2022-16-18-45)-Study-related document.docx
12708Informed consent form    382494-(Uploaded-28-07-2021-22-34-58)-Study-related document.docx



Results publications and other study-related documents

Documents added manually
No documents have been uploaded by study researchers.

Documents added automatically
No additional documents have been identified.