Mobile screening for major depressive disorder in adults from an ethnically and socioeconomically diverse population.

Major depressive disorder (MDD) is a relatively common mental disorder,
affecting more than 264 million people worldwide (James, Abate et al. 2018)
(Liu, He et al. 2020). MDD is therefore the fourth-largest contributor to the
global disease burden and the leading cause of disability worldwide
(Kalibatseva and Leong 2011). Nationally representative samples have indicated
a global prevalence of MDD of 13,3%, which is influenced by a number of
clinical and demographic variables, including age and gender (Abdoli, Salari et
al. 2022). Recurrences following an initial depressive episode are particularly
problematic with an estimated range from 75 to 90% (Gotlib, Goodman et al.
2020). Costs associated with MDD are widely spread: MDD is linked to a reduced
healthy life expectancy and can result in family dysfunction, decreased
professional efficiency, and suicide (Yazdkhasti 2010, Steensma, Loukine et al.
2016, James, Abate et al. 2018). In total, this has resulted in an estimated
economic impact of more than 100 billion euros per year inside the EU union
(Sobocki, Jönsson et al. 2006).

Given the long-term and recurrent nature of MDD as well as its numerous
negative consequences, early detection and treatment of MDD may be
significantly important (Rohde, Lewinsohn et al. 2013). Screening for MDD can
earlier identify individuals requiring support, especially for people who are
unlikely to seek it. A number of validated screening technologies are
available, and screening is already recommended in several settings (Kroenke,
Spitzer et al. 2001) (Siu, Force et al. 2016). Previous studies suggest that
screening may effectively reduce MDD symptoms and significantly increased
complete remission rates (O'Connor, Whitlock et al. 2009).

While the potential for the prevention of MDD using early screening
intervention is substantial (Almeida 2014, Hall and Reynolds-Iii 2014, Park and
Zarate Jr 2019) timely intervention remains a challenge in therapeutic
settings. As a result, there has to be more clarification in screening
protocols regarding the proper screening intervals and follow-up requirements
in order to reduce MDD occurrences. To address this gap, the present study
emphasizes the importance of investigating the feasibility and effectiveness of
screening for MDD by providing a mobile-based screening protocol using a
randomized controlled trial (RCT). We hypothesized that a mobile-based
screening strategy for MDD evaluated in this study protocol will substantially
reduce the burden of MDD over time, improve participants* quality of life, and
minimize disparities associated with MDD.

2. OBJECTIVES

The overarching goal of this proposal is to examine the feasibility and
effectiveness of mobile-based screening for MDD. We combine validated screening
and interventions for MDD with innovative outreach methods to maximize impact.
Thereby, we hypothesize that the mobile-based screening strategy evaluated in
this proposal will substantially reduce the burden of MDD over time, increase
participants* quality of life, and decrease MDD-related disparities.
Specifically, we aim to a) evaluate the feasibility and effectiveness of
mobile-based screening for MDD in a high-risk adult population, b) examine the
impact of subgroup differences in screening participation and effectiveness;
c), develop a personalized screening strategy considering a person*s screening
history and risk characteristics. Our proposal has the potential to inform
practice on the public health impact of screening, including the impact on
health inequalities by sex/gender, racial-ethnic background, employment status,
and other relevant social economic variables. It will also provide detailed
information on the mechanics of MDD across individuals in terms of the duration
and severity of symptoms. This can help inform future screening guidelines on
the appropriate intervals and follow-up criteria for screening. Ultimately, it
may contribute to a lower MDD burden and greater mental health equity in the
population.

Specific Aim 1. Evaluate the feasibility and effectiveness of mobile-based
screening for MDD in a high-risk adult population.
Under this aim, a randomized controlled trial will be conducted to examine the
uptake, health impact and harms-benefits of frequently repeated mobile-based
screening. Eligible participants in the age of 18+ years will be recruited via
a multi-faceted outreach strategy in the municipality of Rotterdam,
particularly from the districts Charlois, Feijenoord (including Kop van Zuid)
and Ijselmonde. We will randomize 1575 eligible respondents across three arms,
in a 1.5:1:1 fashion. The three arms comprise a control arm, a screening arm
with limited participant referral for treatment (after three positive test
scores or suicidal ideation), and a screening arm with standard referral for
participants with moderate-severe symptoms of major depression (single positive
test score). The screening with limited follow-up is vital to get a better
understanding of the natural course of symptoms under usual care, and is
nonexistent in literature to our knowledge. The screening will be conducted
using the 9-item version of the Patient Health Questionnaire (PHQ-9), which has
an estimated sensitivity and specificity of 88% given a cutoff score of 10
(Kroenke, Spitzer et al. 2001). The screening measurement will be solicited
every four weeks during the first 12 months via the Your Research app
(YourResearch 2021). A positive screening test result is considered a PHQ-9
score >=10 (moderate-severe symptoms). Participants with one or three positive
results, respectively (depending on the study arm), will receive an
automatically suggested notification to contact their general practitioner (GP)
The GP has a Praktijk Ondersteuner Huisarts Geestelijke Gezondheidszorg
(POH-GGZ) specialist which is the standard point of access for mental
health-related care. These specialists can refer patients to psychological
therapists in case of further support needs. In addition to the regular
collection of PHQ-9 data, we will ask participants once to provide informed
consent, socio-demographic information and contact information of their GP.
Further, there will be four quality of life measurements (5 questions each, 0,
6, 12, 24 months), and two short screening evaluation surveys (3 questions
each, 6, 12 months). The estimated overall time commitment for participants in
screening is around 120 minutes. This primary aim will demonstrate the
feasibility and effect in QALY gained for repeated mobile-based screening for
MDD. Also, it will provide invaluable information regarding the duration and
severity of MDD symptoms across participants within the screenings arms, on
which the second research question will be based.

Specific Aim 2. Examine participation differences between subgroups and the
impact of screening on disparities.
In this aim, we will use advanced statistical analysis to better understand
differences in screening participation and outcomes between study subgroups.
This aim provides insight into the effect of screening on MDD-related health
disparities. It also provides detailed insight in the mechanics of MDD across
different individuals in terms of the duration and severity of symptoms, and
therefore inform more personalized screening approaches (aim 3).

Specific Aim 3. Develop a personalized screening strategy considering a
person*s screening history and risk characteristics.
This goal will optimize screening intervals and follow-up requirements based on
MDD mechanics observations from aim 2. This objective is built on advanced
knowledge of simulation, machine learning techniques and cost-effectiveness
methodology. To achieve this goal, we will simulate exposure to MDD symptoms in
a hypothetical population using the model for the time exposure to symptoms
from Aim 2. The simulation will include stochastic variation in the duration
and level of MDD symptoms during a 12 month time window, conditional on
participant characteristics and prior MDD symptom history. Screening will be
superimposed with various possible intervals (1 weeks, 2 weeks, * 52 weeks) and
positivity cutoffs (PHQ>=10, *, ), and can be varied depending on patient risk
characteristics. In the simulations, the assumed diagnostic performance of the
test given each positivity cutoff will be derived from Aim 1 data (gold
standard for diagnosis is POH-GGZ referral for treatment), and validated
against literature (Rizopoulos 2022). Assumed effects of follow-up will also be
based on the model for symptom exposure from Aim 2. However, assumed screening
and treatment costs will be as assumed in Aim 1, but varied in sensitivity
analysis. For adherence, we will compare observed rates and with assumptions of
100% and lower rates. An evolutionary algorithm will be used to optimize
screening according to patient characteristics and prior MDD symptom scores
(van Duuren, Ozik et al. 2022). We aim to maximize QALY*s gained given
different criteria for number of GP referrals per QALY gained (harms-benefit
ratio, HBR): Max(INT,POS): Quality of Life, subject to Referrals/QALY < HBR.
(INT=interval; POS=positivity criteria).
Once the algorithm converges on an optimal strategy, we will compare the
benefits and harms from personalized screening with those for uniform screening
(Aim 1). Thereby, we quantify the net benefit from personalization. This goal
will provide more information on how frequently to screen, when to refer
patients for follow-up, and who should be screened more and less intensively.
This has the potential to enhance the balance of screening's risks and
benefits, as well as result equality.

A randomized controlled trial (RCT) involving one thousand and five-hundred and
seventy-five (1575) eligible participants in the age of 18+ will be recruited
within the municipality of Rotterdam-South (particularly the districts
Charlois, Feijenoord and Ijselmonde). Participants will be recruited using
various channels tailored to their cultural background, including direct
approach by representatives of the community, promotion through general
practitioners, local radio, social media and in the public space. RCT will have
3 research arms including one control group, consisting of 675 participants,
and two intervention groups consisting of 450 participants each group. Both
intervention arm will have 4-weekly screening with either lenient follow-up or
screening with stricter follow-up for a time period of one year. Data will be
collected via an app designed by Your Research which runs on Microsoft Azure
server, as the primary of participants response collection. A dedicated back-up
system will serve as a secondary data collection.

This research will make use of a personally designed app (YourResearch 2021)
solicit for digital informed consent as well as socio-demographic information.
Furthermore, both Euro Quality of Life-5D (EQ-5D) and PHQ-9 questionnaires will
be solicited via the application on specific time points. In case participants
within the screening arm exceed a threshold score of 10 points on the PHQ-9
questionnaire, they will be suggested by the application to contact the POH-GGZ
specialist of the GP. Finally, the application features both correct and
necessary certificates and classifications.

The participants are asked to install the app of the study and then fill in the
EQ-5D questionnaire 4 times. The intervention groups are also asked to complete
14x the PHQ-9 and 2x a screening evaluation questionnaire. Each questionnaire
takes less than 5 minutes to complete, which means that the load is considered
low. Although the questions from the PHQ-9 can be confrontational, the
potential health benefits of early detection of depression seem to us to more
than compensate for this.