A Study on Young Individuals' Mental Health in Sweden

In Sweden, reports show an increase in mental health problems of young people, in particular of young women. This increase is poorly understood, calling for more rigorous studies to clarify extent and severity of the problems, but also, to inform practices on how to adequately assess youth mental health.

The purpose of this project is to increase our knowledge about the mental health of young women and men and to establish a valid and comprehensive method for assessing mental health in young individuals that can improve our capacity to promote mental health and address mental health problems. A multimodal data approach is applied, comprising a feasibility study on methods of assessment of mental health in young individuals.

The specific aims of the proposed project are: 1) To determine the extent and course of mental health in young women and men over a two-year period and discern the added value of multimodal data to assess and predict mental health problems. 2) To demonstrate the feasibility of different assessment methods of mental health in young women and men. Establishment of feasibility is done by collecting data on: recruitment rate, adherence and acceptance of study procedures including mode of data collection.

There is a common opinion in Sweden and internationally that mental health problems have increased over the past decades. More research is needed to understand this proposed increase in mental health problems and to clarify if the reports reflect a genuine deterioration in mental health. Existing studies have been criticized for not being able to provide reliable estimates of mental health problems due to heterogeneity in study designs and limitations including lack of use of established, comprehensive measures. The national surveys that indicate a deterioration of mental health in young women in Sweden rely on reports from single questions. Although less respondent burden, the validity and reliability of using single questions to assess concepts such as anxiety and depression can be disputed. Moreover, existing studies provide cross-sectional point estimates rather than prospective time lines, which generally result in more accurate and reliable data.

Traditional methodologies with sole reliance on self-reported data to provide accurate and reliable estimates to better understand mental health of young women may not be the way forward. For instance, metabolomics biomarkers taken from for example blood or urine have gained interest for its ability to screen new and unsuspected pathways involved in psychiatric disorders, and fear conditioning and extinction protocols as well as behavioural testing of e.g. avoidance and distress tolerance have shown promise in detecting and predicting mental health. Combining self-report data with information from such other data sources presents a promising approach, yet the use of multimodal data is an understudied area in mental health research. Moreover, with the ubiquity of smartphones along with advances in technology, it is now possible to use digital devices such as smartphones to assess and track certain behaviours, which makes it possible to address some of the drawbacks of traditional assessment methods. Behavioural real-time data on e.g., participant's symptoms, functions, and social environment in nonclinical settings collected from digital devices holds considerable potential for detecting, understanding and predicting psychiatric conditions, mood disorders, and depressive symptoms.

This feasibility study consists of a prospective cohort using a multimodal data approach. The same theoretical framework, procedure, data collection and measurements apply to both Aim 1 and Aim 2.

Design Three assessment protocols will be evaluated. After baseline assessment, participants will be randomly assigned to a protocol on a 1:1:1 basis. Considering participant dropout after baseline and with an aim to include 750 young individuals per protocol, recruitment will continue until the target number of 200 young individuals per year of birth at baseline (n=3000) is reached. The proposed sample size is based on our primary feasibility outcome: complete adherence of the protocol. Assuming 90% power and alpha=0.05, at least 500 young individuals are needed to adhere to each protocol to detect a 10% statistically significant difference in adherence between protocols, e.g. 60% adherent young individuals assigned to protocol 1 and 2, who complete follow-up, compared with 50% adherent young individuals assigned to protocol 3.

Data collection Data from different modalities will be combined, including both active and passive data. Specifically, data will be collected by survey, smartphone sensor data, at-home blood sampling, behavioural tests, and register data. Most data will be administered via a study-specific app. Written (electronic) consent will be obtained before participation. For instance, a participant may consent to participation with survey data, sensor data, and behavioural tests, but not with blood sample or linkage to register data. Incentives will be offered to participants at completion of an assessment.

In protocol 1, mental health is assessed by survey on three occasions: at baseline and follow-ups at 1 and 2 years later. This design with yearly assessments is congruent with conventional methods of assessment by survey. To investigate the potential benefits as well as risks of more frequent assessments, data will be collected in the same way in protocol 2, but on five occasions: at baseline and 6, 12, 18 and 24 months later. In protocol 3, in addition to yearly assessments by survey, a real-time data capture method referred to as ecological momentary assessments (EMA) will be applied. After the baseline assessment, participants in protocol 3 will asked to complete daily registrations using a selection of self-report measures during a one-month period. The self-report questions will be distributed via the application with about 15-20 questions at a time. This EMA schedule will be repeated three times over the study period.

Data analysis plan The number of young individuals who indicate caseness (according to available scoring guides) of depressive symptoms based on survey responses will be determined to assess prevalence and developmental trajectories of self-reported mental health problems over the two-year follow-up time. Mental health profiles will be mapped with the use of information from the different types of data. The statistical techniques applied for identification of symptom patterns, relationships and determinants of mental health problems and well-being will be suited to complex longitudinal and multimodal data, such as structural equation modelling, generalized linear (mixed) models, and Bayesian modelling. To study time series of symptom occurrence clusters of longitudinally reported symptoms will be obtained by unsupervised clustering analysis.

Based on previous studies on mental health in youth, a point of reference of >20% for baseline recruitment rate will be considered feasible. Dropout analysis will target representativeness at baseline and throughout the phases of the study. At baseline, the distributions of sociodemographic data obtained by register data will be analysed between participants and nonparticipants. For identification of attrition bias, dropout analysis will consider sociodemographic register data but also analyse distributions of health outcomes from previous assessments among participants and dropouts at later stages of the study.

An adherence rate of >50% during the two-year follow-up will be considered feasible, and will be evaluated for each assessment protocol. For the EMA outcomes (protocol 3), the recording and reporting of compliance data will follow the published recommendations, including the measurement of compliance rate as number of prompts answered out of the total number of prompts delivered (compliance of ≥80% feasible) and latency (time period between when participants receive a prompt and when the EMA is answered). Based on previous survey studies on mental health in Sweden, an overall compliance of >75% will be considered feasible. A feasible acceptance rate of behaviour experiments in this heterogeneous, nonclinical population cannot be decided, but our results will clarify acceptance and completion rate for this kind of remotely delivered behavioural experiment. Based on preliminary results of the Mom2B study, an acceptance rate of >70% for access to passive data and 80% to register data will be considered feasible. Furthermore, the completion rate of each self-report measure will be calculated. Instrument guidelines will guide the analyses and interpretation, but if no guideline exists <10% internal missing per question/scale is considered feasible.

Expected outcome The prospective design allows detection of extent and severity of mental health problems in young individuals and will clarify the course of mental health during a two-year period. To clarify the added value of having more frequent assessment points to detect course of mental health is an important outcome of the present study. The results of the EMA assessments will reveal individual day-to-day variability across mental health symptoms and behaviors, expanding the knowledge of the underlying processes and dynamics of mental health. The results will demonstrate the utility of multimodal data for the assessment of mental health and show how outcomes from different data sources are interrelated. This, along with the identification of potential drawbacks of frequent assessments will guide future health assessment strategies. It also allows for data on the predictive value of health outcomes from different methods of assessment. Of particular interest is to demonstrate if multimodal data provide more comprehensive data than single data modality and to clarify if multimodal data improve precision in prediction of mental health problems.

Importantly, the results will demonstrate if an age- and sex-stratified recruitment strategy is appropriate or if selective recruitment methods are needed to obtain a representative population sample. The feasibility of collecting mental health data by self-report measures with 6-month and/or one year follow-up as well as with an EMA approach will be clarified. Results will also show if behavioural testing, passive data, blood sampling are feasible outcome tools to use as proposed.