Systematic review and meta-analyses of studies analysing instructions to authors from 1987 to 2017

The study investigates how journals’ Instructions to Authors (ItAs) have evolved and differ across fields with respect to key research integrity requirements. Reporting practices vary across and within disciplines, and ItAs outline requirements for manuscript preparation, ethical oversight, conflicts of interest, data sharing, and other standards. Despite widespread updates to ItAs, their breadth, changes over time, and variability had not been comprehensively assessed. The authors conducted a systematic review and meta-analyses to synthesize evidence across studies that analyzed ItAs from multiple journals, focusing on six integrity-related topics (authorship, conflicts of interest, data sharing, ethics approval, funding disclosure, and ICMJE URM) and factors potentially explaining heterogeneity (time, country, indexation, impact factor, discipline, sub-discipline).

Prior work suggested substantial differences in reporting practices and policies across journals and disciplines, but comprehensive assessments of ItAs’ content and its evolution were scarce. The included body of literature (1987–2017) pursued varied objectives (e.g., whether specific topics are addressed in ItAs; whether reporting in published articles aligns with ItAs; recommending standards). Evidence indicated suboptimal adherence of published articles to items mentioned in ItAs and that mentioning topics in ItAs often correlates with better, though still suboptimal, reporting in articles. Most existing studies focused on Health Sciences journals, leaving other disciplines less well characterized.

Design: Systematic review and series of meta-analyses following PRISMA guidelines. Protocol: Not preregistered in PROSPERO (non-health outcome), with data and materials available in a Mendeley repository. Eligibility: Studies analyzing ItAs of more than one journal, regardless of topic(s). Information sources: Searches conducted May 1, 2017 in MEDLINE (Ovid), Scopus, and Web of Science without language or time restrictions; Google Scholar query (-allintitle: instructions authors); backward citation searches and known studies. Study selection: Records deduplicated in Rayyan; abstracts screened independently by two reviewers (MM, AJ); disagreements resolved via full-text review. Full texts assessed for eligibility; topics analyzed and percentages of journals addressing each topic extracted. Data extraction: Variables included number of journals analyzed; sampling method; discipline (reclassified into Arts & Humanities, Health, Life, Physical, Social, Multidisciplinary Sciences); sub-discipline; country/region of journals; year ItAs were accessed/analysed; topics assessed; number/percentage of journals addressing each topic; ItA analysis methods; factors assessed for association (e.g., impact factor, indexation, publisher); study objectives. Database indexation recorded as reported. Extraction by MM, checked by AJ. Synthesis: Over 100 topics grouped into 32 major topics, retaining sub-topics separately. Publication volume over time in Crossref, PubMed, Scopus, WoS obtained Dec 4, 2020. Statistical analysis: Time trends in number of ItA studies vs. database publication counts modeled with spline regression (SPSS v24). Meta-analyses of percentages for six topics (authorship, conflicts of interest, data sharing, ethics approval, funding disclosure, ICMJE URM): Comprehensive Meta-Analysis v3, logit transformation, 95% CIs, continuity correction for 0%/100% cases. Random-effects models; if fewer than five studies, fixed-effect model. Heterogeneity: Cochran’s Q and I2. When heterogeneity was considerable, pooled estimates were avoided; instead, mixed-effects subgroup analyses (categorical factors: countries, disciplines, IF categories) and random-effects meta-regressions (time, IF values, discipline, indexation; DerSimonian–Laird). For small-study meta-regressions (n<4), significance inferred by consistency and significant pairwise differences using ECSI software. Pseudo-R2 quantified explained variance. Two-sided tests; p<0.05 primary threshold (noting 0.1 when underpowered). Risk of bias: No established tools for ItA-content studies. Selection biases noted (dominance of Health Sciences journals; heavy reliance on JCR-indexed journals). Reporting limitations common (methods of ItA analysis often unspecified; many studies failed to report the ItA year). All eligible studies included with methodological caveats documented.

• Scope: 153 studies (1987–2017) included; sharp rise in publications after 2002 (growth faster than overall article growth; chi-squared p<0.0001). ItAs were assessed for >100 topics grouped into 32 major topics; most commonly analyzed: Research ethics (n=53, 34%) and Reporting guidelines (n=51, 33%). Median major topics per study: 2 (IQR 1–3). Nearly half (48%) examined associations with at least one factor (15 different factors explored).
• Meta-analyses conducted for six research integrity topics: authorship (n=26 studies), conflicts of interest (n=36), data sharing (n=10), ethics approval (n=31), funding disclosure (n=10), ICMJE URM (n=45). Across topics, between-study heterogeneity was large.
• Six factors explained substantial heterogeneity: time, country, database indexation, impact factor, discipline, sub-discipline. Time trends (examples): • Top/AIM Health Sciences journals: authorship and ethics approval rose from ~40% in 1995 to >70% by 2005. • Health Sciences sub-disciplinary journals: conflicts of interest increased 57% (1995) to 87% (2015); little/no increase after 2000 for URM (~60%), authorship (~65%), funding disclosure (~81%). • URM decreased in AIM Health Sciences journals: 35% (1986) to 5% (2006). • Top journals across disciplines: data sharing ~15% (1992) and conflicts of interest ~16% (1997) to >85% by 2010. • Country-specific increases: Indian Health Sciences journals ethics approval 57% (2004) to 78% (2015); Brazilian 56% (2007) to 83% (2012). Croatian journals (all disciplines) showed increases from ~9% (2013/2014) to >20% (2015) for authorship, conflicts, data sharing. Country differences: • 2014: ~90% of top Health Sciences journals addressed authorship; Chinese 86%, Indian 70%, Croatian 29%. Conflicts of interest: 2010 top journals ~83%; 2014 Indian Health Sciences 89% vs. Croatian 9% (all disciplines). • Highest prevalence in Health Sciences: conflicts of interest in India (89% in 2014), funding disclosure in India (70% in 2008), URM in India (75% in 2014); ethics approval in UK (86% in 2005); authorship in China (86% in 2014). Chinese journals had lowest URM coverage (7% in 2011). Indexation: • Associated with most topics (except data sharing due to lack of comparative data). Example: 1986, funding disclosure in top Health Sciences vs. AIM (47% vs 22%). Ethics approval later higher in AIM (81% in 2006) than top Health Sciences (71% in 2009). • Health Sciences journals in JCR: URM increased 22% (2001) to 77% (2014); AIM decreased 37% (1986) to 5% (2006). • Sub-disciplinary pediatrics in JCR addressed conflicts of interest (78%) more often than DOAJ (63%) (2008–2016); URM ~60% with little indexation difference. Impact factor: • Indications (sub-disciplinary Health Sciences only): conflicts of interest higher in IF≥3 (85%) vs IF<1 (72%) (2008–2016); URM 74% vs 50% respectively; authorship higher in IF 1–2 (61%) vs IF<1 (26%) (2010). Single-study/correlation results were inconsistent elsewhere. Discipline: • Health Sciences more likely to address all six topics than other disciplines (e.g., 2010 WoS: authorship—Health Sciences 59% vs Arts & Humanities 7%). Data limited (1–3 studies per topic). Sub-discipline: • Generally lower coverage in sub-disciplinary Health Sciences than in general/top Health Sciences journals. For authorship (~52% between 1995–2015), funding (~81% between 2000–2015), and URM (~60% between 2008–2016) sub-disciplinary differences were small; larger differences for ethics approval and conflicts of interest. Adherence and effects on reporting: • In 17 studies, journals addressing topics in ItAs tended to have better reporting in published papers, though often <80% adherence. Twelve studies found suboptimal adherence of published papers to stated ItA requirements. Overall: Topic coverage generally increased over the 30-year span, with substantial variation by country, indexation, impact factor, discipline, and sub-discipline.

The review and meta-analyses demonstrate that journals’ ItAs have increasingly addressed key research integrity topics over the past three decades, especially in Health Sciences. The six identified factors—time, country, indexation, impact factor, discipline, and sub-discipline—help explain substantial heterogeneity in coverage. These findings support the premise that ItAs can promote better reporting practices and transparency, though observed adherence in published articles remains suboptimal. Notably, coverage patterns differ by topic; increases are not uniform across topics, fields, or regions (e.g., decreases in URM in AIM journals, divergent country trends), underscoring that the presence of one policy does not imply adoption of others or similar trajectories over time. The dominance of Health Sciences in the evidence base suggests that these journals may be leading in formalizing reporting and integrity standards, potentially informing other disciplines. The results highlight the importance of editorial policies and broader systemic drivers (e.g., regulations, indexation criteria) in shaping ItAs and reporting standards.

This study synthesizes three decades of research on ItAs and quantifies how often six core research integrity topics are addressed, identifying six key factors that explain much of the observed heterogeneity. While coverage generally increased, many sub-disciplinary and regional journals still lag in providing guidance. Stakeholders—publishers, editors, and journals—could strengthen research quality and transparency by updating and implementing robust ItA policies. Future work should: (1) expand analyses beyond the six topics to map broader ItA content; (2) develop comprehensive, public databases tracking journals’ ItA requirements and adherence; (3) evaluate interventions and enforcement mechanisms that improve authors’ compliance; (4) investigate readership and influence of ItAs on author behavior; and (5) explore additional determinants (e.g., editorial leadership, publisher policies, high-profile cases) of policy adoption and change.

• Meta-analyses were restricted to six research integrity topics for feasibility; other topics were not quantitatively synthesized.
• ItAs may not list all enforced practices; conversely, listed practices may not be enforced, and some practices are reported in articles even if not addressed in ItAs, complicating interpretation.
• Associations with country, language, and discipline often relied on few studies (sometimes from the same author/group), limiting generalizability.
• The synthesis treated topic coverage as binary (addressed/not), without granularity on how topics were addressed (e.g., specific authorship criteria).
• Evidence base skewed toward Health Sciences and JCR-indexed journals, introducing selection bias across disciplines and indexers.
• Many studies poorly reported methods and/or the year of the analyzed ItAs, increasing uncertainty and potential reporting bias.