A systematic review published in BMJ Open in 2017 tracked 195 completed PROSPERO-registered reviews and found that the mean time from registration to publication was 67.3 weeks, with an interquartile range of 42 weeks (Borah et al., 2017). That is roughly 15 months from the point when most teams have already been working for weeks on the protocol. The mean team size was 5 people, with a standard deviation of 3. Funded reviews took significantly longer than unfunded ones, with a mean of 42 versus 26 weeks after registration.
If you are a solo PhD student, DNP candidate, or MSN researcher planning to complete a systematic review as part of your thesis or capstone project, those numbers describe a situation significantly different from the one most program timelines assume. A realistic timeline for a systematic review is not 3 months. It is not 6 months in most cases. For researchers who need support at any stage of the process, ScribeLab Writer's systematic review team works with PhD students, MSN and DNP researchers, and faculty teams from protocol development through to manuscript submission.
Quick Answer:
A systematic review takes a mean of 67.3 weeks from PROSPERO registration to publication, requiring a mean team of 5 people (Borah et al., 2017, BMJ Open). In-house, the cost of a single systematic review is approximately $141,194.80 in researcher time (Michelson and Reuter, 2019). The total estimated hours across all stages is a mean of 1,139, distributed across search and retrieval (588 hours), statistical analysis (144 hours), manuscript writing (206 hours), and administration (201 hours) (Allen and Olkin, 1999, JAMA). AI-assisted screening tools can reduce the screening workload by 63.9 to 91.7 percent while maintaining 95 percent recall of relevant studies (Ferdinands et al., 2023). Cochrane MECIR standards require at least two independent reviewers for screening and data extraction. A solo systematic review fails this standard and introduces documented bias.
The Numbers That Define the Systematic Review Workload
Three studies provide the most reliable quantitative picture of systematic review time and cost. Understanding what each one measured and what it did not is essential for using these figures correctly.
Borah et al. (2017): This study analyzed 195 completed PROSPERO-registered reviews using data from the PROSPERO database (registration date to publication date). It found a mean of 67.3 weeks from registration to publication. The clock starts at registration, not at the beginning of protocol development. Most teams spend several weeks developing the protocol before registering, so the total time from project inception to publication is longer than 67.3 weeks. Funded reviews took a mean of 42 weeks after registration; unfunded reviews took a mean of 26 weeks. The mean team size was 5 people with a standard deviation of 3.
Allen and Olkin (1999): This study analyzed 37 meta-analyses from the MetaWorks program at the New England Medical Center and reported total hours per review with a mean of 1,139 hours (median 1,110; range 216 to 2,518). The hours distributed as: search, retrieval, and database development 588 hours; statistical analysis 144 hours; report and manuscript writing 206 hours; and administrative tasks 201 hours. This study is more than 25 years old, and screening hours in particular have changed substantially with AI tools, but the writing and analysis components remain broadly comparable.
Michelson and Reuter (2019): This study calculated the in-house cost of a systematic literature review at $141,194.80, derived as $82,090 per year multiplied by 1.72 years of solo scientist time. This figure represents the researcher's salary cost alone, excluding overhead, software licenses, and librarian time. The study authors disclosed funding from a for-profit company and a potential conflict of interest, so they present this figure as an estimate rather than a definitive benchmark.
These three datasets are the most-cited sources on systematic review time and cost. None of them is a perfect universal estimate. Review complexity, team experience, database scope, and topic density all affect the actual time. Use them as planning anchors, not as guarantees.
Stage 1: Protocol Development and PROSPERO Registration
Protocol development is the stage most researchers underestimate, and most project timelines ignore entirely. Before the search begins, the team must agree on the research question in PICO format, define eligibility criteria precisely enough that two reviewers would reach the same decision on every screened record, specify the databases and grey literature sources, name the risk-of-bias tools for every study design category, and plan the synthesis approach.
A comprehensive protocol document is typically 10 to 20 pages. PROSPERO registration requires completing 23 fields with sufficient detail for CRD staff to assess scope and credibility. Registrations returned for revision most often because the research question or eligibility criteria are too vague. can delay the project by weeks.
Indicative time: 2 to 6 weeks for an experienced team with a clear question; 4 to 12 weeks for a first-time reviewer or a question requiring significant background literature review before the PICO can be finalized. A detailed guide to writing the protocol and completing every PROSPERO field is available in the PROSPERO registration guide.
Stage 2: Search Strategy Development
A reproducible search strategy is not a quick task. Developing a high-quality search string requires identifying controlled vocabulary terms (MeSH in PubMed, CINAHL Subject Headings in CINAHL) alongside free-text synonyms for every PICO component, testing the string to confirm it retrieves known relevant studies, and documenting the full Boolean string for every database in a format that meets PRISMA-S reporting requirements (Rethlefsen et al., 2021).
Working with a medical librarian significantly improves search quality and reduces the risk of systematic errors. The PRESS (Peer Review of Electronic Search Strategies) guideline recommends having a second librarian or information specialist review the search string before it is run (McGowan et al., Journal of Clinical Epidemiology, 2016).
Indicative time: 1 to 3 weeks with a medical librarian; 2 to 6 weeks for a self-directed search with adequate time for testing and refinement. Do not run the search until the protocol is registered. Any search conducted before registration cannot be described as prospective.
Stage 3: Title and Abstract Screening
Title and abstract screening is the stage most responsible for the overall review timeline. When a search across five or six databases retrieves thousands of records, the volume of screening work determines whether the review is completed in 12 months or 24.
Manual screening rates: Experienced reviewers screen approximately 100 titles and abstracts per hour with adequate accuracy. Novice reviewers screen closer to 60 to 80 records per hour. Accuracy degrades after approximately three hours of continuous screening. Most methodologists recommend short sessions of 1 to 2 hours with breaks. For a review retrieving 3,000 records after deduplication, two independent screeners, each working at 80 records per hour, would require approximately 37 hours of screening work each before reconciling disagreements.
Pilot screening: Before the full screening begins, two reviewers independently screen a pilot set of 50 to 100 records and calculate inter-rater agreement. A kappa of 0.6 or above (substantial agreement) confirms that the eligibility criteria are interpreted consistently. Pilot screening takes 2 to 4 hours and prevents much larger time losses from having to re-screen because of inconsistency.
AI-assisted screening: Active-learning tools, particularly ASReview, reduce the number of records requiring manual review while maintaining high recall of relevant studies. Ferdinands and colleagues (2023, Systematic Reviews) tested active-learning models across six datasets and found workload reductions of 63.9 to 91.7 percent at 95 percent recall. In practice, a review that would require 3,000 manual screens may require only 600 to 1,100 screens with AI assistance before reaching the same sensitivity. The Research Screener tool (Scott et al., 2021, Systematic Reviews) reported 60 to 96 percent workload savings with a mean time saving of 12.53 days per review.
Indicative time: 1 to 8 weeks, depending on the number of retrieved records and whether AI assistance is used. Reviews retrieving more than 5,000 records without AI assistance are the most common source of timeline collapse.
Stage 4: Full-Text Retrieval and Screening
Records that pass title and abstract screening require full-text retrieval and a more detailed eligibility assessment. Full-text retrieval involves obtaining the complete paper, which requires inter-library loan or direct author contact for articles not available through your institution's subscriptions. Retrieval delays are common and can add days or weeks to this stage.
Full-text screening requires reading each paper in sufficient detail to confirm it meets all eligibility criteria. The typical inclusion rate from full-text screening to final inclusion is 20 to 50 percent, meaning that for every 100 full texts retrieved, 50 to 80 are ultimately excluded for specific, documented reasons. PRISMA 2020 now requires listing all excluded studies with reasons, which must be tracked systematically throughout this stage.
Indicative time: 1 to 4 weeks for most reviews, depending on the number of full texts to retrieve and assess.
Stage 5: Data Extraction
Data extraction translates the findings of each included study into a structured format for synthesis. The variables extracted depend on the review question: intervention type, population characteristics, sample sizes, outcomes, effect sizes with measures of variability, measurement tools, follow-up duration, funding sources, and risk-of-bias information.
Cochrane MECIR standards require at least two people working independently to extract outcome data. Single-reviewer extraction misses transcription errors and introduces selection bias in which data are extracted from complex tables and figures.
Data extraction time per study rises with the complexity of the extraction form. A simple form with 15 to 20 variables takes approximately 20 to 45 minutes per study. A complex form covering multiple outcomes, multiple time points, and intervention subgroups takes 60 to 90 minutes per study. Piloting the extraction form on three to five studies before full extraction confirms the form is workable and catches ambiguous fields.
Indicative time: For a review with 20 included studies and a standard extraction form, expect 10 to 25 hours of extraction work per reviewer, plus time for reconciling disagreements. For reviews with 50 or more included studies, data extraction is often the second-largest time commitment after screening.
Stage 6: Risk-of-Bias Assessment
Risk-of-bias assessment applies a validated appraisal tool to each included study. The time per study depends on which tool is used and how many outcome-level assessments are required.
Minozzi and colleagues (2020, Journal of Clinical Epidemiology) measured RoB 2 application time at a mean of 28 minutes per study outcome before a structured guidance document was implemented. RoB 2 is applied per outcome, not per study. A trial with three reported outcomes requires three separate assessments. ROBINS-I V2 covers seven domains per study and requires more time per assessment than RoB 2 for most reviewers. NOS (Newcastle-Ottawa Scale), while simpler, has documented reliability limitations and requires a judgment call on the comparability domain that often generates reviewer disagreement.
The critical appraisal guide covers every major risk-of-bias tool in detail, including decision rules, signaling questions, and common application errors.
Indicative time: 30 to 60 minutes per study for RoB 2 (depending on outcome count); 45 to 90 minutes per study for ROBINS-I V2. Total appraisal time for 20 studies across two independent reviewers is typically 20 to 60 hours.
Stage 7: Meta-Analysis and Statistical Synthesis
Meta-analysis time depends primarily on the number of outcomes, the number of subgroup analyses, the complexity of the statistical models, and whether meta-regression is conducted. Allen and Olkin (1999) reported a mean of 144 hours for statistical analysis across 37 meta-analyses.
Selecting the statistical model, running the analyses, generating forest plots, calculating heterogeneity statistics (I², tau², prediction intervals), and conducting sensitivity analyses in RevMan, R, or Stata is a task that requires methodological knowledge of meta-analytic models as well as statistical software proficiency. Reviews requiring network meta-analysis (comparing more than two interventions simultaneously) require specialist software and expertise beyond the scope of most standard systematic review training.
Indicative time: 2 to 8 weeks for a standard meta-analysis with one to three primary outcomes and pre-specified subgroup and sensitivity analyses.
Stage 8: Manuscript Writing
Allen and Olkin (1999) reported a mean of 206 hours for report and manuscript writing across 37 meta-analyses. Cochrane planning guidance cites writing and internal review as taking 1 to 11 months, depending on the complexity of the review and the number of authors contributing to and reviewing drafts.
A systematic review manuscript includes: abstract, introduction, methods section (covering every PRISMA 2020 item), results section (PRISMA flow diagram, characteristics of included studies, risk-of-bias summary, synthesis results), discussion (including GRADE certainty), and references. Most journal submission systems require the PRISMA 2020 checklist as a required supplementary file.
Indicative time: 4 to 12 weeks for a first complete draft; add 2 to 6 weeks for internal revision across co-authors.
Stage 9: Submission, Peer Review, and Revision
Beller and colleagues (2013, PMC3674908) analyzed the time from last search to publication for a sample of systematic reviews and found substantial delays between search completion and publication. Peer review timelines vary by journal, but most clinical journals take 2 to 6 months from submission to first decision. Revision cycles add further time. Major revisions typically require 1 to 3 months to address.
Desk rejection at high-impact journals (The Lancet, JAMA, BMJ, NEJM) occurs for approximately 70 to 90 percent of submissions before they reach peer review. A cascade strategy ranking journals by impact factor and specificity to the review topic, and planning for resubmission, is essential for managing this stage efficiently.
Indicative time: 3 to 12 months from first submission to acceptance, depending on journal tier and revision requirements.
Table 1: Systematic Review Stage-by-Stage Time Breakdown
Stage | Indicative Timeframe | Biggest Time Driver | Tools That Help |
|---|---|---|---|
1. Protocol and PROSPERO registration | 2 to 12 weeks | Finalizing the PICO question and writing precise eligibility criteria. PROSPERO returns for vague fields, add weeks. | PROSPERO (crd.york.ac.uk). OSF for scoping reviews (osf.io). |
2. Search strategy development | 1 to 6 weeks | Building and testing the Boolean string across multiple databases with controlled vocabulary terms. Faster with a medical librarian. | PRESS guideline (McGowan et al., 2016) for peer review of search. CINAHL, PubMed MeSH tools. |
3. Title and abstract screening | 1 to 8 weeks | Volume of retrieved records. Manual rate: ~80 records per hour. AI-assisted: 64 to 92% workload reduction at 95% recall (Ferdinands et al., 2023). | ASReview (free, active learning). Rayyan (free, web-based). Covidence (paid, Cochrane standard). |
4. Full-text retrieval and screening | 1 to 4 weeks | Inter-library loan delays for articles not available through institutional subscriptions. 20 to 50% of full texts are typically excluded at this stage. | Unpaywall browser extension. Open Access Button. Direct author contact for unavailable papers. |
5. Data extraction | 1 to 6 weeks | Complexity of the extraction form. Simple forms: 20 to 45 min/study. Complex forms: 60 to 90 min/study. Dual independent extraction doubles the total effort. | Covidence extraction forms. SRDR+ (free, AHRQ). DistillerSR (paid, enterprise). |
6. Risk-of-bias assessment | 1 to 4 weeks | RoB 2: ~28 minutes per outcome (Minozzi et al., 2020). ROBINS-I V2: longer per study (7 domains). Dual assessment doubles total effort. | riskofbias.info (RoB 2, ROBINS-I V2 templates). robvis (mcguinlu.github.io) for traffic-light plots. |
7. Meta-analysis and statistical synthesis | 2 to 8 weeks | Number of outcomes, subgroup analyses, and meta-regression models. Mean 144 hours across 37 meta-analyses (Allen and Olkin, 1999). | RevMan (Cochrane, free). R with meta/metafor packages (free). Stata (paid). CMA (paid, beginner-friendly). |
8. Manuscript writing | 4 to 12 weeks | Methods section (every PRISMA 2020 item), results section (flow diagram, evidence tables, forest plots), GRADE Summary of Findings tables. Mean 206 hours (Allen and Olkin, 1999). | GRADEpro GDT (gradepro.org) for Summary of Findings tables. PRISMA 2020 checklist (EQUATOR network). |
9. Submission and peer review | 3 to 12 months | Desk rejection at high-impact journals (70 to 90% of submissions). Peer review cycles and major revision requests. Prepare a cascade of 3 to 5 target journals in advance. | JANE (journalauthornameestimator.biosemantics.org) for journal selection. PRISMA 2020 checklist for submission compliance. |
The Solo Reviewer Problem
Cochrane MECIR standards (Methodological Expectations of Cochrane Intervention Reviews) require at least two people working independently for eligibility determination at both the title/abstract and full-text screening stages, and at least two people extracting outcome data independently. PRISMA 2020 items 8 and 9 require authors to report how many reviewers screened records and extracted data, and whether they worked independently.
These requirements are methodological standards, not administrative preferences. Single-reviewer screening introduces selection bias because any individual reviewer consistently applies their own interpretation of ambiguous eligibility cases rather than having those cases resolved through discussion. Single-reviewer extraction misses errors in data transcription and allows the reviewer's prior beliefs about which studies support the hypothesis to influence which data are extracted subtly.
The practical consequence for solo researchers is stark: to conduct a systematic review that meets Cochrane and PRISMA 2020 standards, you need a second independent reviewer for screening and extraction. The Borah data confirms this. The mean team size for a completed review was 5 people. Even the minimum defensible approach (two reviewers) requires a collaborator for the most time-intensive stages.
For PhD students whose supervisors are listed as co-authors, the supervisor typically fulfills the second reviewer role for a sample of screened records or all extracted data. For DNP and MSN students whose programs do not provide a second reviewer, this is a genuine methodological challenge that should be addressed in the project design before the search begins, not after.
Working on a systematic review alone and falling behind on your timeline? |
|---|
Cochrane MECIR standards require at least two independent reviewers for screening and data extraction, and the mean systematic review team is 5 people. Solo reviewers face both a methodological problem and a workload problem that most academic timelines do not account for. ScribeLab Writer's systematic review team provides independent second-reviewer support for screening, data extraction, and risk-of-bias assessment, helping solo researchers meet Cochrane and PRISMA 2020 standards without needing to recruit a co-author. |
How AI Tools Change the Timeline
AI-assisted screening tools do not eliminate reviewer work. They reorganize it. Active learning tools like ASReview prioritize the order in which records are presented for screening, placing the most likely relevant records first. As the reviewer includes and excludes records, the model updates its ranking. The result is that most relevant records surface early in the screening process, allowing the reviewer to reach a stopping criterion before manually reviewing every record.
Ferdinands and colleagues (2023) tested six AI models across multiple systematic review datasets. They found that active learning reduced the proportion of records requiring manual screening to between 8.3 and 36.1 percent of the total, while maintaining recall of relevant studies at 95 percent. The implication: a review that would require 5,000 manual screens may require only 420 to 1,800 screens before the stopping criterion is met.
The time savings are real but come with methodological obligations. The AI model used, the stopping criterion applied, and the recall rate achieved must be disclosed in the methods section. The RAISE framework (Responsible use of AI in evidence SynthEsis) provides guidance on what must be reported when AI tools are used in systematic reviews.
Stages AI tools can accelerate:
Title and abstract screening (most significant impact: 64 to 92 percent workload reduction)
Full-text screening prioritization (emerging capability)
Data extraction (emerging. requires careful validation and disclosure)
Deduplication across databases
Stages AI tools do not replace:
Protocol development and PICO formulation
Search strategy development (requires librarian expertise and controlled vocabulary knowledge)
Risk-of-bias assessment (requires judgment on each signaling question)
Synthesis and manuscript writing
What Makes a Review Take Longer Than Average
The Borah data show an interquartile range of 42 weeks around the mean of 67.3 weeks, meaning reviews routinely deviate substantially from the average in both directions. The factors that most consistently increase review time are:
Broad research question: A narrow PICO question retrieves a manageable number of records. A broad question retrieves tens of thousands. Every additional 1,000 records adds approximately 10 to 15 hours of screening work at manual screening rates.
Multiple databases with poor deduplication: Cross-database duplicates can represent 30 to 40 percent of retrieved records. Poor deduplication before screening inflates the screening workload substantially.
Complex eligibility criteria: The more judgment calls required at screening, the higher the inter-rater disagreement rate, and the more time required for reconciliation.
Multiple outcomes requiring separate appraisal: RoB 2 is applied per outcome, not per study. A review with five primary outcomes requires five appraisal assessments per study.
Author non-response for missing data: When studies report inadequate data for meta-analysis, contacting authors adds weeks or months with no guarantee of a response.
Team size and coordination: Larger teams reduce individual workload but increase coordination overhead. Version control, task assignment, and reviewer reconciliation across five or more people requires project management that smaller teams avoid.
Table 2: Factors That Affect Systematic Review Timeline
Factor | Impact on Timeline | Mitigation |
|---|---|---|
Broad research question | High. More records retrieved, more screening work, more studies included, more extraction and appraisal time. | Tighten the PICO at the protocol stage. Every additional population or intervention category multiplies the workload downstream. |
No AI screening tool used | High for large retrieval sets. Manual screening of 5,000 records at 80 records per hour requires ~62 hours per reviewer. | Use ASReview or Rayyan for AI-assisted prioritization. Disclose the tool, stopping criterion, and recall rate in the methods section. |
Solo reviewer | Medium to high. Single-reviewer screening and extraction are faster but fail MECIR standards and introduce bias. The reconciliation stage is also eliminated, removing a quality check. | Identify a second reviewer before the search begins. A co-author, supervisor, or external reviewer can fulfill this role for specific stages. |
Multiple outcomes requiring RoB 2 per outcome | High. A review with 5 primary outcomes and 20 included RCTs requires 100 separate RoB 2 assessments at ~28 min each. | Pre-specify a limited number of primary outcomes in the protocol. Consider whether all secondary outcomes require separate appraisal or can share a study-level judgment. |
Poor cross-database deduplication | Medium. Cross-database duplicates can represent 30 to 40% of retrieved records. Screening duplicates wastes time with no benefit. | Deduplicate before importing to the screening tool. Use Zotero's duplicate detection, Covidence's AutoDedup, or the Systematic Review Accelerator's Deduplicator. |
Author non-response for missing data | Medium. Waiting for unreported data delays extraction and synthesis by weeks or months, with no guarantee of a response. | Set a maximum waiting period (commonly 4 to 6 weeks) before proceeding without the data. Document the attempt in the methods section. |
High-impact journal target | High. Desk rejection rates of 70 to 90% at top-tier journals mean most manuscripts require 2 to 4 submission cycles before acceptance. Each resubmission adds weeks of reformatting and waiting time. | Build a 3 to 5-journal cascade before the first submission. Prepare reformatted versions for second- and third-choice journals in advance so resubmission takes days rather than weeks. |
Building a Realistic Project Plan
A Gantt-style project plan with stage-level milestones is the most effective tool for managing a systematic review timeline. The stages are largely sequential. The search cannot be run before the protocol is registered, screening cannot begin before the search is complete, and extraction cannot begin before the included studies are finalized, with some overlap possible between stages that do not depend on each other.
Key planning principles:
Allow a buffer of 20 to 30 percent above your estimated time for every stage. Reviews almost never finish ahead of schedule, and frequently fall behind due to factors outside the team's control.
Set a search lock date before starting extraction. Once the search is complete, the database of retrieved records is fixed. Any additional searching after extraction has begun constitutes an amendment that must be documented.
Plan the submission cascade before submitting. Identify three to five target journals in priority order, confirm each one accepts systematic reviews, and confirm the formatting requirements for each. A rejection that requires reformatting for the next journal adds weeks to the publication timeline.
Build the PRISMA 2020 checklist completion into the writing stage, not as an afterthought at submission. Tracking PRISMA items throughout writing is significantly faster than retrospectively confirming compliance.
Timeline Expectations Across Review Contexts
Cochrane reviews: Cochrane planning guidance describes the typical review as requiring approximately 18 months from protocol to publication. Cochrane reviews undergo editorial oversight, peer review of both the protocol and the full review, and production formatting that extends beyond standard journal timelines.
Non-Cochrane journal reviews: The Borah mean of 67.3 weeks from registration applies most directly to this category. High-impact journal submission (The Lancet, JAMA, BMJ) adds further time through peer review and revision cycles.
DNP scholarly projects: US DNP programs vary in what they require. Some require a full systematic review with meta-analysis; others accept a structured literature review or integrative review. Confirm with your program exactly which evidence synthesis type is required before committing to a full systematic review timeline.
Rapid reviews: Rapid reviews deliberately compress the systematic review process by restricting the search scope and accepting single-reviewer screening for some stages. They are completed in 1 to 6 months but carry explicit methodological limitations that must be stated in the final report.
International research (UAE, Saudi Arabia, UK, Australia): The timeline demands are consistent across these contexts because all four markets follow the Cochrane and PRISMA 2020 methodology. Institutional factors (library access, supervisor availability, and ethics review requirements) are the primary sources of timeline variation between countries.
Frequently Asked Questions About Systematic Review Timelines
Can one person complete a systematic review?
Technically, yes, but a solo systematic review fails Cochrane MECIR standards, which require at least two independent reviewers for screening and data extraction. PRISMA 2020 items 8 and 9 require transparent reporting of the number of reviewers and whether they worked independently. A solo review that is honest about this limitation faces scrutiny from peer reviewers and editors. Most methodologists consider a solo review a meaningful methodological weakness that should be disclosed as a limitation in the final manuscript.
How does the Borah 2017 timeline compare to my program's expectations?
Many nursing, public health, and social work programs have capstone timelines of one semester (3 to 4 months) for a systematic review. The Borah data show that the mean time from PROSPERO registration to publication alone is 67.3 weeks. Even if we account for the time from registration to search completion being much shorter than full publication time, most semester timelines are unrealistic for a true systematic review. Confirm with your supervisor whether a structured literature review or an integrative review would satisfy the program requirements with a more realistic timeline.
What is the fastest a systematic review can realistically be completed?
The minimum in the Borah dataset was 6 weeks from registration to publication for the fastest reviews. These are almost certainly rapid reviews, reviews of very narrow topics with few included studies, or reviews with pre-existing data. A typical single-topic intervention review with 10 to 30 included studies and a meta-analysis would rarely be completed in under 6 months from protocol development to submission, even with an efficient team and AI-assisted screening.
Does using AI for screening compromise the review's credibility?
Not if disclosed and conducted appropriately. AI-assisted screening is an accepted methodology when the tool, stopping criterion, and recall rate are reported transparently. Journals, including Systematic Reviews and Evidence-Based Medicine, have published guidance on AI disclosure requirements for systematic reviews. The RAISE framework provides the current standard for what must be reported.
How many databases does a typical systematic review search?
Most systematic reviews in health and medicine search a minimum of three to five databases: MEDLINE/PubMed, Embase, and one discipline-specific database (CINAHL for nursing, PsycINFO for psychology and mental health, Cochrane Central for controlled trials). The specific databases should be justified in the methods section; adding more databases increases retrieval volume and screening time, but also reduces the risk of missing relevant evidence.
What proportion of a systematic review's time is spent on writing?
Based on Allen and Olkin (1999), approximately 18 percent of total hours go to report and manuscript writing (206 of 1,139 mean hours). This is often the stage researchers underestimate relative to the analysis stages, particularly the methods section, which must document every PRISMA 2020 item in sufficient detail for the review to be reproducible.
You May Also Find Useful
Planning Your Review With Realistic Expectations
The Borah data are not intended to discourage researchers from undertaking systematic reviews. It is intended to replace the wishful timelines that result in abandoned reviews, incomplete manuscripts, and the 46 percent non-publication rate that Runjic and colleagues documented in PROSPERO-registered reviews.
A systematic review completed well with a registered protocol, a reproducible search, dual independent screening and extraction, a valid risk-of-bias assessment, and transparent reporting is one of the most impactful outputs in evidence-based health research. The way to protect that impact is to plan for the actual time it takes, not the time it would be convenient for it to take.
A systematic review that begins without a realistic stage-by-stage plan consistently stalls at the screening or data extraction stage, when the gap between projected and actual workload becomes unmanageable, and the choice becomes compressing the methodology or abandoning the review entirely. ScribeLab Writer's systematic review team, led by credentialed researchers with published systematic reviews in the biomedical literature, provides support at any individual stage, screening, extraction, risk-of-bias assessment, statistical synthesis, or across the full review from protocol development to submission-ready manuscript. Submit your protocol details and the stage where you need support through the enquiry form, and a member of the team will respond within 2-4 hours.
