Study eligibility criteria
Types of studies
For the initial search of risk and protective factors for depression in adolescence in HICs and LMICs, studies using longitudinal prospective designs will be identified. A second search will focus specifically on LMICs to examine cross-sectional study designs due to the lack of longitudinal studies evidenced in these settings. Finally, when considering the relationship between biological, psychological, and contextual factors and adolescent depression, we will include longitudinal and cross-sectional studies and intervention and prevention trials in order to capture any biological or contextual factors that may be evidenced while considering the gap in longitudinal study designs conducted in LMICs. Trials or experimental designs will only be included when examining biological factors in order to expand the pool of potential studies. Discussion papers, letters, editorials, case studies or case series, and qualitative studies without a quantitative element will be excluded.
Type of participants and settings
Both sexes in adolescent populations will be included in this review. Included studies will require a depression outcome (as defined by a categorical DSM/ICD diagnosis, clinical diagnosis, medical records, structured interview, or self-report measure). Eligible populations will be young people in high-, middle-, and low-income countries who are under the age of 25 years at study baseline and have been evaluated for depression with at least one subsequent time point of at least 6 months. The LMIC-specific analysis will also include young people living in LMICs who are under the age of 25 years who have been evaluated for the presence of depression or depressive symptoms. The biological factors’ objective will include adolescents in high-, middle-, and low-income countries who are under the age of 25 years and were evaluated for depression and for biological markers of interest, as well as psychological or contextual markers associated with depression. Given the high rates of psychiatric co-morbidity, populations with depression plus other psychiatric conditions (including substance abuse) will be included for the risk and protecte factors part of the review. Studies limited only to specific medical subpopulations (e.g., only young people with HIV, diabetes, intellectual disabilities) that do not include comparative populations will be excluded. Any longitudinal studies with an initial prospective time point of under 6 months will also be excluded. Medication or other treatment status will not be an exclusion criterion, but information will be reported on these whenever available.
Type of exposures
The following factors will be evaluated for association with depression among young people:
-
Contextual factors associated with an increased risk for developing major depression during adolescence (this review will include demographic, economic, family, neighborhood, environmental, and social and cultural factors; in addition, we will explore life events and traumatic exposures);
-
Biological markers associated with an increased risk for developing major depression during adolescence. This review will focus on studies that address the following: hypothalamic-pituitary-adrenal axis (HPA axis), inflammation, endocannabinoids, vitamins, polyunsaturated fatty acids (PUFAs), hormones, neurotrophic factor, neurotransmitters, telomere/gene length, neuroplasticity, and gene expression (including mRNA quantification);
-
Brain-related abnormalities in individuals at risk for major depression during adolescence (this review will focus on magnetic resonance imaging studies, both structural and functional, assessing children/adolescents at risk);
-
Psychological factors associated with increased or decreased risk for developing major depression during adolescence (this review will include factors such as self-esteem, locus of control, cognitive biases, emotional regulation, reward responsivity, and learned helplessness).
Outcomes
The primary outcome will be development of depression among young people with a cutoff of 25 years of age or younger and also allowing for studies that include earlier endpoints of depression development, in any population, i.e., high-, middle-, and low-income countries. The primary outcome measures of depression may include self-reports, adult-informant reports, structured observations in clinical settings, and clinical records.
Timing of outcome assessment
Eligible studies should include assessment of depression during at least one time period between the ages of 10 and 24 years and a follow-up time point at least 6 months later.
Search strategy for identifying relevant studies
We will search the following electronic databases to identify potentially relevant studies for our review: MEDLINE (via Ovid), PsycINFO, Cochrane Database of Systematic Reviews, Web of Science (Core Collection), Lilacs, African Journals Online, and Global Health. In addition to these electronic databases, we also plan to identify relevant studies by reviewing reference lists of eligible studies and review articles and by contacting experts and authors of eligible studies by email. Relevant systematic reviews from the database outputs will be hand-searched for any eligible studies which may have been missed. Only published research in academic journals will be used for objectives 1 and 3 of this review (see below). Searches will be conducted in English, though publication language will not be a limiter. The Boolean operator “OR” will be used to find articles with one or more search terms and synonyms and “AND” to combine the different concepts for relevant articles. Where applicable, MeSH subheadings will be used, and “keywords” where relevant for given databases. For objective 2 (see below), which will use data from LMIC, we will expand our search to grey literature. Some studies of adolescents in LMIC are presented in reports of nongovernmental organizations, and these findings may not be reported in the academic literature. Therefore, we will review reports from international and national nongovernmental organizations and will attempt to identify additional literature through listservs used by mental health psychosocial practitioners working in nongovernmental organizations.
We will use truncation and wildcards to account for UK and US spelling and terminology and abbreviation. The concept combinations will vary with objective (see
Appendix 1), but the three main concepts combined will include Population AND Exposure AND Outcome, for example, adolescence AND risk/protective factors AND depression. Years of publication will not be used to limit the search output, but extraction will begin in descending chronological order, assessed at the first 10 years for sufficient robust evidence, then proceed to the next 10 years or more if deemed necessary to capture representative data.
Searches will be re-run before the date of final analyses to identify recent publications in the field.
Objective 1—identify risk and protective factors associated with the development of depression among adolescents and young people in populations evaluated for at least 6 months (i.e., longitudinal samples).
Sample MEDLINE (Ovid) search:
Adolescent/ OR “adolesc*” AND Depression/ OR Depressive disorder/ AND Longitudinal Studies/ OR “longitudinal”
Objective 2—identify risk and protective factors associated with depression among adolescents and young people in LMICs.
Sample MEDLINE (Ovid) search:
Adolescent/ OR “adolesc*” AND Depression/ OR Depressive disorder/ AND Developing Countries/ OR “developing countr*” OR World Bank List (as keywords)
Objective 3—identify risk and protective factors associated with biological markers, brain related abnormalities, and context among adolescents and young people.
Sample MEDLINE (Ovid) search:
Adolescent/ OR “adolesc*” AND Depression/ OR Depressive disorder/ AND “HPA axis” OR Pituitary-Adrenal System
See
Appendix 1 for full MEDLINE search, Objective 1.
Data collection
Selection of studies
A charting form will be used to identify studies to be included in the review according to the inclusion and exclusion criteria. Two review authors will independently review titles and abstracts of studies identified through the search strategy and those from additional resources (e.g., contacting authors and searching reference lists of included papers, which will be done through hand searches, Web of Science citation records, or Scopus when appropriate) to determine whether studies potentially meet the inclusion criteria. To assess for study eligibility, two reviewers will independently review the full publication texts of studies identified through title and abstract review. They will record justification for exclusion for each excluded study. A third author will resolve disagreements between the two reviewers.
Retrieval of studies
The full text of the selected articles will be retrieved. When a full article is not available through the database or university library services, it will be searched through ResearchGate and Google Scholar, or a request to share the full article will be sent by email to the corresponding author. After the request has been sent, a reminder will be sent after a week. If there is no response within 2 weeks from the reminder, this document will be excluded from the systematic review.
Data extraction and management/coding
Extraction fields will include, but not be limited to, article information (publication year, journal), study characteristics (country, urban vs. rural setting, funding), population characteristics (age at enrolment, current age, number of longitudinal assessments), depression assessment (depression criteria, duration of depression, assessment tool, co-morbidity), risk factors (poverty, trauma, family history), protective factors (education, social capital), assessment tool (instrument names and psychometric properties for depression, risk, and protective factors), and biological variables (interleukins, cortisol, galvanic skin response, neuroimaging). Two authors will independently extract data from the published reports of included studies using a standardized, electronic form (e.g., in Microsoft Excel), which will be developed and pre-piloted to assure interrater agreement. We will attempt to collect missing information by contacting the corresponding authors via email.
Assessment of risk of bias in included studies
Two reviewers will independently assess the risk of bias in included studies by evaluating the following domains:
-
Recruitment and sampling strategy
-
Follow-up strategy, retention, and selective attrition
-
Completeness of outcome data
-
Validity and other psychometrics of depression outcomes and other tools used
-
Selective outcome reporting
-
Other sources of bias related to context, procedures, and reporting
We will assess the quality of observational studies using the Systematic Assessment of Quality in Observational Research (SAQOR) [
35] and the adapted SAQOR Cultural Psychiatry Epidemiology (SAQOR-CPE) [
36]. If any intervention studies are included, we will limit the analyses to be pre-treatment findings or control groups. We will apply the SAQOR-CPE to the samples of interest in the intervention studies. Quality level will be recorded as high, moderate, low, or very low risk of bias for each domain, and bias rating will be compared between the two author reviewers. A third author will resolve discrepancies. Emails to authors will be used to obtain additional information to reduce sources of bias and adjust risk.
Subgroup analysis and assessment of heterogeneity
If possible, we will conduct subgroup analyses based on HICs vs. LMICs, by age of onset, co-morbidities, outcome type (e.g., self-report, clinical interview, medical records), geographic region, and population settings (primary care, non-psychiatric specialty services, communities, religious centers, schools). Some included studies may report onset of depression after 25 years of age for some participants. Therefore, we will potentially conduct subgroup analyses by age group of onset: those who have developed depression before the age of 25 and those who have developed it after the age of 25 for studies that follow participants into adulthood. Subgroup analyses will be conducted in Comprehensive Meta-Analysis, including analysis of statistical difference among subgroups. We will assess heterogeneity using the chi-squared test,
I-squared statistic, and prediction interval [
40]. We will evaluate for potential sources of heterogeneity. We will address heterogeneity by using random-effects meta-analysis models and sensitivity analyses. We will use the approach of referring to “low, moderate, and high” heterogeneity at the
I2 thresholds of 25%, 50%, and 75%, with the caveat that quantification of heterogeneity will be considered only one marker of variability across studies [
41] and the concerns raised by Cuijpers [
42] that even point estimates of
I2 may have wide confidence intervals extending into the ranges of high heterogeneity.
Presentation of results
We will present the results of our systematic review through a narrative description addressing the study’s objectives and through charts and tables as recommended in the Cochrane Handbook (Section 11) [
43]. First, we will tabulate data from our search process into a flowchart indicating the number of studies identified and excluded at key steps, according to the PRISMA guidelines [
25]. Second, we will summarize information on individual studies in a “Characteristics of Included Studies” table. Third, we will present “Data and Analyses” tables and forest plots to present outcome data from individual studies and from any meta-analyses. Fourth, we will present a “Summary of Findings” table to summarize key data available, magnitude of effect of the risk and protective factors studied, and quality of evidence addressing the key study objectives.