Study design, reporting, and registration

This scoping review will be conducted in accordance with JBI methodology [32–34] and reported using the PRISMA Extension for Scoping Reviews [35]. Protocol reporting follows PRISMA-P [36,37], as required by PLOS ONE for scoping-review Study Protocols; the completed checklist is provided as S1 Checklist. The protocol is registered at OSF Registries (https://doi.org/10.17605/OSF.IO/NPWTV). Because the submitted registration is immutable, current working materials, future review data, and versioned supporting files will be maintained in the associated public OSF project (https://osf.io/rav8m/). Substantive departures from the registered plan will be dated and justified in S3 Appendix and submitted as a formal OSF registration update where applicable.

As of July 13, 2026, the MEDLINE strategy is a draft and PRESS review has not been completed. No formal bibliographic database search, screening, data charting, critical appraisal, or synthesis has begun. Preparatory registry searches undertaken to contextualize novelty do not constitute the formal review search.

Objectives and analytic orientation

The review addresses descriptive mapping questions and does not posit a confirmatory hypothesis. All analyses are exploratory and will describe the extent, characteristics, and distribution of eligible evidence. No null-hypothesis testing, meta-analysis, effect pooling, or certainty-of-evidence grading is planned.

Conceptual organization

The seven modules are coordinated organizational structures rather than components of one causal theory. Definitions and classification will preserve distinctions among recognized diagnoses, conditions for further study, and problematic-use constructs. Measurement findings will be organized using COSMIN-defined properties and the per-use evidence tiers specified below [38,39]. Determinants will be mapped to I-PACE categories without treating I-PACE as proof of causal direction [40–42]. Intervention descriptions will use TIDieR [43], implementation outcomes will use Proctor’s taxonomy [44], and equity-related reporting will use PROGRESS-Plus and SAGER [45–49]. Table 1 limits each framework to its intended role.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Coordinated module architecture.

https://doi.org/10.1371/journal.pone.0348860.t001

Review questions

Seven linked questions are presented in Table 2. Questions 1–4 address definitions and classification, measurement, epidemiology, and determinants; Questions 5–6 concern harms, interventions, and implementation; Question 7 is cross-cutting.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Review questions.

https://doi.org/10.1371/journal.pone.0348860.t002

Eligibility criteria

Context, dates, and languages.

No geographic region, country-income category, or clinical, community, school, university, household, or online setting will be excluded. “All regions” denotes the absence of a geographic eligibility restriction; it does not remove the language limitation below.

Databases will be searched from inception. No publication-date restriction will be applied in the search or during screening. Dates associated with diagnostic recognition, manual editions, instrument development, or policy adoption will be charted as historical descriptors and may be used for stratified presentation; they will not determine eligibility or appear as exclusion reasons in Fig 1.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA 2020 flow diagram template for the planned review.

This blank two-lane template will be populated after searching, screening, and study selection. ProQuest Dissertations & Theses Global and WHO Global Index Medicus are classified as databases; Google Scholar, websites and organizations, and backward and forward citation searching are classified as other methods. Full-text exclusion reasons reflect the prespecified eligibility criteria. Publication date is not an exclusion criterion. Adapted from the PRISMA 2020 flow diagram [59], distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license (https://creativecommons.org/licenses/by/4.0/).

https://doi.org/10.1371/journal.pone.0348860.g001

Records in English, Spanish, French, German, Portuguese, Arabic, and Hebrew are eligible. No database-level language filter will be applied. Records in other languages will be excluded only after publication language has been identified during screening, and language-specific exclusion counts will be reported. Machine translation evidence from particular language pairs cannot establish reliability for a different language or task [51,52]; therefore, the human-verification procedure below applies to every final non-English decision and finding.

Evidence types.

Eligible primary evidence includes quantitative, qualitative, mixed-methods, intervention, implementation, and psychometric studies. Systematic reviews, meta-analyses, and scoping reviews are eligible as secondary evidence but will be charted and synthesized separately from primary studies. Official classifications, guidelines, and consensus statements are eligible only when retrieved through the prespecified database, grey-literature, or citation-searching pathways; they will provide contextual information and will not be combined with research outputs. Ad hoc surveillance briefs and standalone policy or implementation documents are not formal sources of evidence.

Editorials, letters without original data, opinion pieces, narrative commentaries, book chapters without original data, and conference abstracts are excluded. Full peer-reviewed conference papers or proceedings are eligible when they otherwise meet the criteria.

Table 3 consolidates the boundary rules used at screening and charting.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Eligibility and boundary rules.

https://doi.org/10.1371/journal.pone.0348860.t003

Information sources and search strategy

The bibliographic and grey-literature database search will include MEDLINE via Ovid, PsycINFO via Ovid, CINAHL via EBSCO, Scopus, Web of Science Core Collection, the Cochrane Library, IEEE Xplore, ProQuest Dissertations & Theses Global, and WHO Global Index Medicus. Prespecified supplementary methods will include Google Scholar, a targeted search limited to six organizations, and backward and forward citation searching. The organizational pathway is limited to WHO/ICD, the American Psychiatric Association and PsychiatryOnline, UNICEF, UNESCO, the International Telecommunication Union, and NICE. For each listed domain, six exact phrase queries will be run as logged-out site-restricted web searches; all results will be screened when 50 or fewer are returned and otherwise the first 50 in relevance order will be screened. The complete draft MEDLINE strategy and all supplementary-source procedures are reported in S1 Appendix and will be deposited in the associated OSF project before screening [33,53].

The principal database strategy will combine target-construct and instrument terms with population terms. The MEDLINE draft combines the current Internet Addiction Disorder heading and legacy Behavior, Addictive indexing with relevant technology headings and broad free-text variants for addiction, dependence, dysregulation, maladaptive or pathological use, overuse, and problematic or compulsive use. The population block includes the un-exploded Adult heading and the adult* free-text term to protect recall for the 25–29 comparison stratum; some records involving older adults will therefore be retrieved and resolved during screening. Measurement, prevalence, determinant, harm, intervention, qualitative, implementation, and equity terms will not form a mandatory third concept block because requiring them could omit evidence relevant to the seven questions. Searches will run from database inception without date or language limits, consistent with sensitivity-oriented search guidance [54].

Three prespecified Google Scholar queries will address gaming, social-media, and smartphone constructs. For each query, the first 200 results in relevance order will be screened, following Haddaway et al. [55]. Searches will be conducted without an authenticated Google account; the exact query, date and time, displayed result count, browser and session conditions, number screened, and retained citation details will be logged. Incognito mode will be used to reduce account-history effects, not claimed to eliminate personalization. Google Scholar is supplementary and will not replace database or targeted organizational searching.

Backward reference-list searching and forward cited-reference searching will be conducted for all included sources using TARCiS terminology [56,57]. Dates, platforms, and results will be logged. The registry searches described in the Introduction will be updated immediately before formal search execution and before resubmission if the package is delayed; exact strings and limitations will be retained in S1 Appendix.

Before execution, the MEDLINE strategy will undergo PRESS peer review by an independent information specialist [58]. Resulting changes will be recorded in S3 Appendix, incorporated into every database translation, and included in the OSF update. No claim is made that a librarian or information specialist has already reviewed the strategy.

Record management and study selection

Search results will be imported into Covidence and assigned stable source identifiers. Exact duplicates matched on stable identifiers may be collapsed automatically, with each action retained in the audit log. Every non-exact or near-duplicate cluster will require human confirmation; ambiguous clusters will be checked by a second reviewer. No automated function will exclude a record on substantive eligibility grounds.

Two reviewers will independently screen every title and abstract. Before full screening, they will assess a 50-record calibration set supplemented as needed to represent target constructs and evidence types. Full screening will begin after percent agreement reaches at least 80%; Cohen’s kappa will be reported when calculable. If the threshold is not met, eligibility guidance will be clarified and tested on 50 new records. Any substantive rule change will be versioned and recorded in S3 Appendix.

Two reviewers will then independently assess full texts. A ten-report full-text calibration exercise will precede full assessment. Disagreements at either stage will be resolved through discussion and, if necessary, adjudication by a third reviewer. One primary exclusion reason will be recorded for each excluded full text using the hierarchy in Table 3. Selection will be reported in the PRISMA 2020 flow diagram (Fig 1) [59].

Human verification of non-English evidence

The team has documented access to source-language-proficient reviewers for Spanish, French, German, Portuguese, Arabic, and Hebrew. If that capacity becomes unavailable before or during the review, the affected work will pause until coverage is restored or the protocol is formally amended.

Machine translation may be used to prepare a working rendering, but no eligibility decision, exclusion reason, charted value, interpretation, or quotation will rest solely on machine output. For every record in an eligible non-English language, a source-language-proficient person will check the original for each final title-and-abstract and full-text decision. A second reviewer will assess the human-verified rendering when making the independent decision.

For charting, the source-language reviewer will verify construct terminology, eligibility-relevant wording, direction of association, numerical values, uncertainty measures, qualitative findings, and any quotation against the original. The second reviewer will chart from the verified rendering and source locator. The translation log will record language, verifier, stated proficiency, machine tool and version if used, date, fields checked, discrepancies, and corrections. English-language evidence will undergo the same dual human screening and charting process but will not require translation verification.

Data-charting forms and fields

Charting will use Pollock’s dual-table structure: Table A defines variables, accepted values, decision rules, critical fields, and hypothetical examples; Table B contains populated records [32,60]. S2 Appendix provides complete Table A forms for all seven modules before calibration. No module form is deferred for development during full charting.

Each retrieved document will be assigned unique source and report identifiers. Primary research will also receive a study identifier and, for each distinct analyzed sample or wave, a sample identifier; secondary reviews will receive a review identifier. A dataset-family identifier and any other applicable identifiers will be assigned only when relevant and otherwise coded N/A. Core fields include citation, country, setting, design, sample size, age distribution, sample- or result-level age-based evidence classification, sex and gender reporting, source language, source level, target construct, instrument and version, measurement tier, and applicable module assignments. Each charted quantitative or qualitative finding will receive a unique result identifier. Result fields include construct, instrument use and tier, population stratum, primary and linked modules, result role, temporality, direction, estimate, uncertainty, adjustment status, source locator, whether the field is designated critical, translation provenance, and overlap links.

The initial charting forms contain 32 core fields, including a source-level appraisal method. Before calibration, the initial core form and all seven module-specific forms will be version-locked. Applicable module forms contain 12 fields for definitions and classification, 22 for measurement, 15 for epidemiology, 13 for determinants, 12 for harms, 26 for interventions and implementation, and 14 for equity; the separate qualitative-integration matrix and citation matrix linking included reviews to their cited primary studies contain nine and six fields, respectively. A source completes one core record and only its applicable module and result rows, so workload grows with the number of reported samples, instruments, and results; no source requires completion of every field. During calibration, reviewers will log minutes per source–module pair, applicable and missing field counts, reconciliation time, and reviewer assignment. These observations will determine workload allocation and any timeline revision before full charting. If adequate dual-reviewer capacity cannot be sustained, charting will pause and any staged output or scope change will be documented as a protocol amendment before work resumes.

Charting calibration and full charting

Before full charting, two reviewers will independently chart at least five study-module pairs for each of the seven modules and at least ten unique reports overall. The same report may contribute to more than one module. If fewer than five eligible pairs are available for a module, every available pair will be used and the small denominator reported.

The calibration denominator includes every jointly applicable structured categorical field. NR, NA, and UNCLEAR are categorical decisions: disagreement between either of these codes and another code or value, or between two different missingness codes, counts as disagreement. A field is excluded from the denominator only when both reviewers independently select NA and reconciled applicability confirms that the field does not apply. The gate requires at least 90% agreement overall and at least 80% agreement within each module and for every critical field with at least one applicable comparison. Continuous numerical values and source locators will be compared exactly and reconciled separately rather than included in the categorical denominator. Cohen’s kappa will be reported for suitable categorical fields when calculable but will not replace the percent-agreement gates.

If a module or critical field fails, its guidance and hypothetical examples will be revised and tested on five new study–module pairs, or all remaining new pairs if fewer than five exist. Earlier records affected by a rule change will be re-charted under the new codebook version. A critical field with no applicable paired ratings will be recorded as not estimable, not passed. Reviewers will add eligible calibration pairs until the field has an applicable comparison; if every available pair is jointly and correctly NA, a third reviewer will confirm the applicability decision, the field will be labeled not applicable in calibration, and calibration will reopen before further charting when the first applicable source appears. Full charting will begin only after every estimable gate passes and each field with no applicable paired ratings has followed this prespecified deferral procedure.

After calibration, two reviewers will independently chart every included source into its applicable modules. Disagreements will be resolved by consensus and, when necessary, third-reviewer adjudication.

Result-role classification and cross-module overlap

Classification will occur at the result level rather than once per construct or study.

• A factor measured before and modeled as predicting a later target construct will be classified as a determinant.
• A target construct measured before and modeled as predicting a later outcome will be classified as a harm.
• In a cross-sectional model, an association in which a factor is modeled as the predictor and the target construct as the dependent variable will be assigned to the determinants view.
• In a cross-sectional model, an association in which the target construct is modeled as the predictor and another variable as the dependent variable will be assigned to the harms view.
• An undirected cross-sectional association will be stored once, cross-referenced in the determinants and harms views, and counted once.
• Reciprocal or cross-lagged paths will receive separate result identifiers for each reported direction.
• A prespecified post-intervention endpoint will be classified as effectiveness.
• An unintended intervention outcome will be classified as an adverse event.
• Acceptability, adoption, appropriateness, feasibility, fidelity, cost, penetration, and sustainability will remain implementation outcomes.

A result may be displayed in more than one cross-referenced view but will have one count in the underlying dataset. Directed results will have one primary module; a genuinely undirected concurrent result may have a null primary module and approved linked-module views. This structure prevents, for example, a depression-gaming association from being duplicated merely because it is relevant to determinants and harms.

Multiple reports, secondary evidence, and overlap

Reports arising from the same study will share a study identifier; reports based on the same participant dataset will also share a dataset-family identifier when applicable. When reports contain the same estimate, it will be charted once from the most complete source and linked to companion reports. Companion publications will be retained for unique outcomes, time points, subgroups, or methods.

Primary studies, systematic reviews or meta-analyses, scoping reviews, and contextual guidelines or consensus statements will be stored at separate source levels. Review conclusions will not be pooled, vote-counted, or merged with primary-study results. A primary study cited by an included review will enter the primary dataset only if independently retrieved and eligible under this protocol; eligibility will never be inherited from a review.

A citation matrix will link each included review to the primary studies identified in its reference list. Secondary reviews may suggest a potential gap, but a gap will be reported only after the direct primary-evidence map is checked. Scoping reviews will remain separate from systematic reviews in charting and appraisal.

Qualitative and mixed-methods synthesis

Qualitative findings will undergo descriptive framework and content analysis. Initial deductive parent codes will correspond to the seven review questions; reviewers may add inductive subcodes when a finding is not adequately represented. The coding unit will be an author-reported finding, linked to its source locator and any supporting participant quotation. Participant quotations and author interpretations will be stored in separate fields.

Two reviewers will independently code all eligible qualitative findings. A five-study qualitative calibration set, or all available studies if fewer than five exist, will require at least 80% agreement. If the threshold is not met, the codebook will be revised and tested on five new studies, or all remaining new studies if fewer than five exist. Codebook versions, consensus decisions, and retrospective recoding will be documented.

Contradictory, disconfirming, and context-specific findings will be retained. Qualitative findings will be presented beside relevant quantitative mappings but will not be meta-aggregated or treated as pooled evidence. The review will not claim thematic saturation, interpret frequency as importance, or apply GRADE-CERQual.

Targeted critical appraisal

Two reviewers will independently assign and apply exactly one source-level appraisal method from Table 4, using the current JBI tool suite where specified [63], with disagreements resolved by consensus or third-reviewer adjudication. Method assignment first uses source level; primary empirical sources are then classified by their stated primary aim and the design supporting the primary result. Explicitly integrated mixed-methods sources use MMAT. A multipurpose source receives one method and is not appraised repeatedly merely because it reports secondary analyses or outcomes. A source whose design remains unresolved during screening may be recorded as pending classification, but it must be assigned a listed method before appraisal. If an eligible source cannot be mapped, appraisal pauses and a prospective protocol amendment must add an appropriate method before that source or any later source is appraised; no ad hoc closest-tool substitution is permitted. Appraisal is intended to describe methodological limitations; it will not determine eligibility, generate composite scores, weight findings, rank instruments, or support certainty-of-evidence grades. Results will be reported by checklist item or domain.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Source-level, non-exclusionary appraisal map.

https://doi.org/10.1371/journal.pone.0348860.t004

Contacting source authors

Source authors will be contacted only when missing information concerns an eligibility-critical age subgroup, the identity of the target construct or instrument and cutoff, or a primary estimate essential to a prespecified module output. Routine unreported fields will be coded NR without contact.

One request will be sent to the corresponding author, followed by one reminder after four weeks if necessary [67]. Charting will continue while a response is pending. Contacts, replies, corrections, and unresolved fields will be logged in the associated OSF project; a nonresponse will not cause exclusion. Public logs will omit personal contact information.

Synthesis and presentation

Synthesis will be descriptive and organized by review question, construct, age-based evidence classification, measurement tier, and source level. No meta-analysis is planned. Numerical findings will be reported as provided, with uncertainty measures when available, and will not be combined across incompatible definitions, instruments, thresholds, or age strata.

Definitions will be summarized in a terminology and classification map. Measurement findings will be presented in an instrument-by-property matrix that preserves language, population, tier, and appraisal. Epidemiological findings will be separated by ascertainment method, instrument, cutoff, tier, and age stratum. Determinants and harms will be mapped by temporality and result role. Intervention findings will separate effectiveness, adverse events, and implementation. Equity reporting will be mapped across PROGRESS-Plus and SAGER fields.

Qualitative findings will be integrated through side-by-side tables and narrative comparison. Secondary reviews will have separate summary tables. The evidence-gap map will identify combinations of constructs, populations, settings, methods, and outcomes for which eligible evidence is sparse or absent; because of language and retrieval boundaries, absence will be described as absence from eligible evidence rather than proof that no evidence exists. Planned outputs are summarized in Table 5.

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Planned presentation outputs.

https://doi.org/10.1371/journal.pone.0348860.t005

Data management and availability

Versioned search files, deduplication records, screening decisions, exclusion reasons, codebooks, translation logs, charting tables, appraisal records, links between included reviews and cited primary studies, and analytic scripts will be maintained in the associated OSF project. Personally identifying correspondence data will not be made public; any public contact ledger will contain only non-identifying process fields.

Use of artificial intelligence tools

OpenAI Codex (OpenAI; accessed July 2026) assisted with language editing, cross-file consistency checks, reference-metadata checking. Human authors checked and approved all affected text, references, tables, and supporting files. Codex did not make eligibility, appraisal, synthesis, or interpretive decisions and did not generate study data. The authors retain responsibility for accuracy, attribution, interpretations, limitations, and conclusions.

Ethics and safety

The review will use published or publicly accessible documentary evidence and will not recruit participants, administer interventions, or access individual-level confidential data. Ethics approval, informed consent, sample-size calculation, randomization, blinding, participant safety monitoring, and qualitative saturation are therefore not applicable. The review will preserve accurate attribution and will not reproduce identifying participant quotations beyond what source authors published.

Study status and timeline

As of July 13, 2026, the search strategy is drafted, PRESS review is pending, and no formal database search, screening, charting, appraisal, or synthesis has begun. PRESS review is projected for July-August 2026, search execution for September 2026, screening and charting through April 2027, and completion of synthesis and domain-focused reports by July 2027. These dates are projections and will be updated transparently if the schedule changes.

Appropriateness and expected contribution

A scoping review is suited to these questions because it can map divergent terminology, measurement practices, study designs, outcomes, and populations without forcing them into one pooled estimate [31–34]. The review will not determine that all included constructs are equivalent disorders. Instead, it will preserve their nosological differences and show how each has been defined, measured, and studied. This mapping can identify focused questions for later systematic reviews without presenting descriptive breadth as certainty about effects.

The coordinated modules may show where apparent differences arise from measurement choice rather than population differences, where temporal evidence supports or fails to support directional interpretation, and where intervention evidence lacks implementation or equity information. These are anticipated uses of the map, not protocol-stage findings.

Strengths and limitations

Strengths include one shared sensitive search, explicit age and per-use measurement-tier rules, complete module-specific charting forms, result-level direction and overlap controls, dual human screening and charting, source-language verification, separation of primary and secondary evidence, and a non-exclusionary source-level appraisal method. The staged module structure is intended to make the broad scope operational while retaining one transparent selection pipeline.

The breadth of seven questions may still limit detail within individual modules, and it creates a substantial charting burden. Calibration, assignment to relevant modules, restricted author contact, and versioned guidance reduce but do not eliminate this risk. Heterogeneous definitions, cutoffs, and measurement properties will restrict comparison; the review will not pool estimates or recommend instruments solely from frequency of use.

Language eligibility is limited to English, Spanish, French, German, Portuguese, Arabic, and Hebrew. Source-language verification should reduce translation error within those languages, but exclusion of Chinese, Japanese, Korean, Russian, and other languages may under-represent culturally specific constructs, instruments, determinants, interventions, and policy contexts. Meta-epidemiological findings about the effect of English-language restrictions on pooled estimates [68–70] cannot establish conceptual neutrality in this evolving field. Language exclusions will be counted, and affected outputs will be framed as maps of eligible evidence rather than complete representations of every regional literature.

The strict age policy protects age-specific interpretation but will reclassify some commonly reported samples spanning ages 18–29, or broader ranges, as contextual evidence. Critical appraisal will describe limitations without excluding or weighting studies, so mapped availability should not be interpreted as proof of evidential strength. Automated exact-duplicate detection may introduce error, although human confirmation of non-exact clusters and retained audit logs provide safeguards.

Dissemination and practical implications

Findings may be reported in two domain-focused manuscripts: definitions and classification, measurement, epidemiology, and determinants (Questions 1–4); and harms, interventions, implementation, and equity (Questions 5–7). Both will derive from the same search and selection process, identify distinct result sets, and cross-reference shared methods without duplicating result counts.

For practice, the measurement map may inform selection of instruments appropriate to a construct, language, and population without substituting for clinical judgment. For service and policy planning, instrument-specific epidemiological and equity maps may clarify where available estimates are comparable. They may also show where surveillance gaps remain. For researchers, the terminology, directionality, intervention, implementation, and gap maps may identify questions suitable for focused systematic reviews or new primary studies. Outputs will be deposited openly in the associated OSF project and shared through peer-reviewed publications and relevant clinical, public-health, education, and youth-health forums.

Protocol amendments

Material departures from the registration will be dated, justified, and recorded in S3 Appendix. Working materials were deposited in RAV8M, which will also hold future review data and code. Any subsequent substantive departure will receive a dated amendment and, where applicable, a further formal registration update.