The DOI matches its Crossref record; 21 of the 21 references it cites were found in a registry or an index; 6 of the 21 sources its claims cite were checked for support.
Better duplicate detection for systematic reviewers: evaluation of Systematic Review Assistant-Deduplication Module
Open accessAbstract
BACKGROUND: A major problem arising from searching across bibliographic databases is the retrieval of duplicate citations. Removing such duplicates is an essential task to ensure systematic reviewers do not waste time screening the same citation multiple times. Although reference management software use algorithms to remove duplicate records, this is only partially successful and necessitates removing the remaining duplicates manually. This time-consuming task leads to wasted resources. We sought to evaluate the effectiveness of a newly developed deduplication program against EndNote. METHODS: A literature search of 1,988 citations was manually inspected and duplicate citations identified and coded to create a benchmark dataset. The Systematic Review Assistant-Deduplication Module (SRA-DM) was iteratively developed and tested using the benchmark dataset and compared with EndNote's default one step auto-deduplication process matching on ('author', 'year', 'title'). The accuracy of deduplication was reported by calculating the sensitivity and specificity. Further validation tests, with three additional benchmarked literature searches comprising a total of 4,563 citations were performed to determine the reliability of the SRA-DM algorithm. RESULTS: The sensitivity (84%) and specificity (100%) of the SRA-DM was superior to EndNote (sensitivity 51%, specificity 99.83%). Validation testing on three additional biomedical literature searches demonstrated that SRA-DM consistently achieved higher sensitivity than EndNote (90% vs 63%), (84% vs 73%) and (84% vs 64%). Furthermore, the specificity of SRA-DM was 100%, whereas the specificity of EndNote was imperfect (average 99.75%) with some unique records wrongly assigned as duplicates. Overall, there was a 42.86% increase in the number of duplicates records detected with SRA-DM compared with EndNote auto-deduplication. CONCLUSIONS: The Systematic Review Assistant-Deduplication Module offers users a reliable program to remove duplicate records with greater sensitivity and specificity than EndNote. This application will save researchers and information specialists time and avoid research waste. The deduplication program is freely available online.
Overview
CoChat Summary
TL;DR
The Systematic Review Assistant-Deduplication Module (SRA-DM) was iteratively developed as a heuristic deduplication algorithm and compared with EndNote X6 on biomedical search results. It identified substantially more duplicate citations than EndNote while maintaining perfect specificity in validation, though it did not detect all duplicates.
Key findings
- On the 1,988-citation benchmark, the fourth SRA-DM iteration achieved 84.4% sensitivity and 100% specificity: 674 true positives, 1,189 true negatives, 0 false positives, and 125 false negatives. EndNote achieved 51.2% sensitivity and 99.8% specificity: 410 true positives, 2 false positives, 1,185 true negatives, and 391 false negatives.
- In three validation datasets, SRA-DM was consistently more sensitive than EndNote: cytology screening (n = 1,856) 90% vs 63%; stroke (n = 1,292) 84% vs 73%; haematology (n = 1,415) 84% vs 64%. SRA-DM specificity was 100% with no false positives, whereas EndNote’s average specificity was 99.7%.
- Across all four bibliographic searches, the average percentage increase in duplicates detected by SRA-DM compared with EndNote was 42.8% (reported as 42.86% in the discussion).
- During development, reported sensitivities across iterations were 75.0%, 95.7%, 68.0%, and 84.4%; the discussion describes the most sensitive iteration as 96% with specificity 99.75% and three false positives.
Methods
- The SRA-DM algorithm was developed iteratively in 2013 at the Bond University Centre for Research in Evidence-Based Practice using a heuristic-based approach, with each version tested against a benchmark dataset of 1,988 citations from a 29 July 2013 search for surgical and non-surgical management of pleural empyema across six databases: MEDLINE-Ovid, EMBASE-Elsevier, CENTRAL-Cochrane Library, CINAHL-Ebasco, LILACS-Bireme, and PubMed-NLM.
- The benchmark was created by importing citations into EndNote, sorting by author, manually coding records as unique or duplicate, reordering by article title, and reinspecting for further duplicates. SRA-DM’s fourth iteration was compared with EndNote X6’s default one-step auto-deduplication using matching criteria of author, year, and title with “ignore spacing and punctuation” ticked.
- Validation used three additional datasets from cytology screening tests (n = 1,856), stroke (n = 1,292), and haematology (n = 1,415). Results were coded against the benchmark as true positive, false positive, true negative, or false negative, and sensitivity and specificity were calculated.
Limitations
- The paper states that the deduplication program was developed for biomedical database citations and was not tested on other bibliographic records such as books and governmental reports, so it may not perform as well with other bibliographies. It also states that some duplicates may have been missed during manual benchmarking, although the database was screened twice and cross-checked.
- The paper states that SRA-DM was compared against the typical default EndNote deduplication setting, while some information specialists use additional steps such as multi-stage screening or updating incomplete citations. It also notes that no software can currently detect all duplicate records, and undetected SRA-DM duplicates occurred due to missing page numbers, author name variance, or extraneous OVID MEDLINE title information.
- Evident from the text: the final algorithm was chosen with lower sensitivity (84%) than a more sensitive iteration (96%) to preserve 100% specificity, so residual duplicates remain; the validation datasets come from specific medical topics and searches, and the comparison is limited to EndNote X6 default rather than other reference managers or enhanced EndNote workflows.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
6 Valid · 0 Not valid · 10 Not checked · 41 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“Typically, auto-deduplication is only partially successful [ 18 ]”
What the sources say
- Supportedcites [18] Qi X, Yang M, Ren W, Jia J, Wang J, Han G, Fan D: Find duplicates among the PubMed, EMBASE, and Cochrane library databases in systematic review.PLoS One 2013, 8:e71838. 10.1371/journal.pone.0071838 source
The Discussion says about 10% of records remained duplicates after auto-searching and recommends combining auto- and hand-searching. The Conclusions explicitly state that auto-searching alone was inadequate, especially for type II duplicates. (LLM, conf=0.90)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“A recent study by Qi [ 18 ] examined the content of undetected duplicate records in EndNote and found that errors often occurred due to missing or wrong data in the fields, especially for records retrieved from EMBASE database.”
What the sources say
- Supportedcites [18] Qi X, Yang M, Ren W, Jia J, Wang J, Han G, Fan D: Find duplicates among the PubMed, EMBASE, and Cochrane library databases in systematic review.PLoS One 2013, 8:e71838. 10.1371/journal.pone.0071838 source
The Methods (Auto-search duplicates; Difference between index and redundant papers of type I duplicates) describe using EndNote to find duplicates and classify missing or wrongly recorded reference fields as errors. The Results (Portal vein thrombosis literatures; Budd-Chiari syndrome literatures) report substantially more wrong information in duplicates found by hand-searching and identify EMBASE as the source with the highest proportion of errors in several fields. (LLM, conf=0.91)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The first iteration of the deduplication algorithm achieved 75.0% sensitivity and 99.9% specificity (Table 3 ).”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“SRA-DM algorithm was consistently more sensitive (Table 4 ) at detecting duplicates than EndNote [cytology screening: 90% vs 63%; stroke: 84% vs 73% and haematology: 84% vs 64%]”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In contrast, the average specificity of EndNote was lower (99.7).”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
16 more claims
Create a free account to see all 57 claims and every source verdict.
- Probably not a substantial claim. 17
- It may state two separate claims as one. 26
- The quote only partly states this claim. 1
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“When used alone, EndNote’s sensitivity for deduplication ranges from 51 to 57% [ 7 , 10 ]”
From Reducing systematic review burden using Deduklick: a novel, automated, reliable, and explainable deduplication algorithm to foster medical research (2022) · cites this paper as [7]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“and this leads to enormous increases in the cost of bridging the synthesis gap 28 .”
From Software support for environmental evidence synthesis (2018) · cites this paper as [28]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“When searching in multiple databases and additional information sources, as is necessary for a systematic review, authors often employ a variety of techniques to reduce the number of duplicates within their results prior to screening [135 -138].”
From PRISMA-S: an extension to the PRISMA Statement for Reporting Literature Searches in Systematic Reviews (2021) · cites this paper as [137]
In the citing paper
“This will prevent the potential bias that self-evaluation can bring [23] as well as provide outside users with confidence in the tools that are created [24].”
From Making progress with the automation of systematic reviews: principles of the International Collaboration for the Automation of Systematic Reviews (ICASR) (2018) · cites this paper as [24]
In the citing paper
“Automatic reference deduplication algorithms have shown acceptable performance results [27], but inconsistent evaluation results remain across datasets and in the way their performance is assessed across the literature [27][28][29].”
From A narrative review of recent tools and innovations toward automating living systematic reviews and evidence syntheses (2023) · cites this paper as [27]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“674 citations as duplicates (TP), 1,189 citations as unique records (TN), no false positives occurred (100% specificity) and only 125 duplicate records were undetected (FN).”
“Of these, 410 were correctly identified as duplicates (TP) and two were incorrectly designated as duplicates (FP), and 1,185 citations were correctly identified as unique records (TN) and 391 duplicate citations were undetected (FN).”
Citations
Cites
Cochrane Handbook for Systematic Reviews of Interventions: Cochrane Book Series
(2008) · doi:10.1002/9780470712184 · Cited by 25,048
Increasing value and reducing waste in research design, conduct, and analysis
Ioannidis, J. P. A., Greenland, S., Hlatky, M. A., Khoury, M. J., Macleod, M. R., Moher, D., Schulz, K. F., & Tibshirani, R. (2014) · The Lancet · doi:10.1016/s0140-6736(13)62227-8 · Cited by 1,597
How to increase value and reduce waste when research priorities are set
Chalmers, I., Bracken, M. B., Djulbegovic, B., Garattini, S., Grant, J., Gülmezoglu, A. M., Howells, D. W., Ioannidis, J. P. A., & Oliver, S. (2014) · The Lancet · doi:10.1016/s0140-6736(13)62229-1 · Cited by 1,436
Reducing waste from incomplete or unusable reports of biomedical research
Glasziou, P., Altman, D. G., Bossuyt, P., Boutron, I., Clarke, M., Julious, S., Michie, S., Moher, D., & Wager, E. (2014) · The Lancet · doi:10.1016/s0140-6736(13)62228-x · Cited by 1,319 · cited 1×
“Waste in research occurs for several methodological, legislative and reporting reasons [19-22].”
BackgroundSelf-citationIncreasing value and reducing waste: addressing inaccessible research
Chan, A.-W., Song, F., Vickers, A., Jefferson, T., Dickersin, K., Gøtzsche, P. C., Krumholz, H. M., Ghersi, D., & van der Worp, H. B. (2014) · The Lancet · doi:10.1016/s0140-6736(13)62296-5 · Cited by 887
Royle, P., & Milne, R. (2003) · International Journal of Technology Assessment in Health Care · doi:10.1017/s0266462303000552 · Cited by 693 · cited 1×
“More recently, the overlap in MEDLINE and EMBASE was found to be 79% [13] based on trials that had been included in 66 Cochrane systematic reviews.”
BackgroundSemi-automated screening of biomedical citations for systematic reviews
Wallace, B. C., Trikalinos, T. A., Lau, J., Brodley, C., & Schmid, C. H. (2010) · BMC Bioinformatics · doi:10.1186/1471-2105-11-55 · Cited by 351 · cited 1×
“Typically of the thousands of citations retrieved for a systematic review search, over 90% are excluded on the basis of title and abstract screening [5].”
BackgroundUnderstanding PubMed(R) user search behavior through log analysis
Islamaj Dogan, R., Murray, G. C., Neveol, A., & Lu, Z. (2009) · Database · doi:10.1093/database/bap018 · Cited by 265 · cited 1×
“Background Identifying trials for systematic reviews is time consuming: the average retrieval from a PubMed search produces 17,284 citations [1].”
BackgroundFind Duplicates among the PubMed, EMBASE, and Cochrane Library Databases in Systematic Review
Qi, X., Yang, M., Ren, W., Jia, J., Wang, J., Han, G., & Fan, D. (2013) · PLoS ONE · doi:10.1371/journal.pone.0071838 · Cited by 98 · cited 3×
“Typically, auto-deduplication is only partially successful [18], and the onerous task of manually sifting and removing the remaining duplicates rests with reviewers or information specialists.”
BackgroundMethodInfluentialLefebvre, C., Eisinga, A., McDonald, S., & Paul, N. (2008) · Emerging Themes in Epidemiology · doi:10.1186/1742-7622-5-13 · Cited by 93 · cited 1×
“However, the methodological details of trials are often inadequately described by authors in the titles or abstracts, and not all records contain an abstract [4].”
Background
Research works printed but not linked: 1
- Systematic review assistant - deduplication module. [ http://crebp-sra.com ]
Cited by
Rethlefsen, M. L., Kirtley, S., Waffenschmidt, S., Ayala, A. P., Moher, D., Page, M. J., & Koffel, J. (2019) · doi:10.31219/osf.io/sfc38 · Cited by 3,696
Rethlefsen, M. L., Kirtley, S., Waffenschmidt, S., Ayala, A. P., Moher, D., Page, M. J., Koffel, J. B., PRISMA-S Group, Blunt, H., Brigham, T., et al. (2021) · Systematic Reviews · doi:10.1186/s13643-020-01542-z · Cited by 3,714
De-duplication of database search results for systematic reviews in EndNote
Bramer, W. M., Giustini, D., de Jonge, G. B., Holland, L., & Bekhuis, T. (2016) · Journal of the Medical Library Association : JMLA · doi:10.3163/1536-5050.104.3.014 · Cited by 2,103
Dual u-net with resnet encoder for segmentation of medical images
Syed, Q. N., & Ismail, A. R. (2022) · The International Islamic University Malaysia Repository (The International Islamic University Malaysia) · Cited by 1,142
O'Dea, R. E., Lagisz, M., Jennions, M. D., Koricheva, J., Noble, D. W. A., Parker, T. H., Gurevitch, J., Page, M. J., Stewart, G., Moher, D., et al. (2021) · Biological Reviews · doi:10.1111/brv.12721 · Cited by 697
“…can be conducted using a single database, but it is preferable to use more than one database for the final search (Bramer et al., 2018) (requiring duplicated studies to be removed prior to study selection, for which software is available; Rathbone et al., 2015a; Westgate, 2019; Muka et al., 2020).”
Method
Metadata
- Authors
- Rathbone, J., Carter, M., Hoffmann, T., & Glasziou, P.
- Affiliations
- Bond University (AU)
- Corresponding author
- Rathbone, John
- Venue
- Systematic Reviews
- Published
- 2015-01-14
- Volume
- 4
- Issue
- 1
- Pages
- 6-6
- ISSN
- 2046-4053 (ISSN-L)
- PMC
- PMC4320616
- OpenAlex
- W1561556558
- Cited by
- 206
- References
- 21
- Citations by year
- 2026: 10 · 2025: 6 · 2024: 39 · 2023: 29 · 2022: 38 · 2021: 32 · 2020: 11 · 2019: 10 · 2018: 12 · 2017: 4 · 2016: 12 · 2015: 3
- FWCI
- 1.70
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Systematic Reviews (publishedVersion, cc-by)
- Europe PMC (PubMed Central) (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- Bond University Research Portal (Bond University) (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Topics
- Social Sciences › Decision Sciences › Statistics, Probability and Uncertainty › Meta-analysis and systematic reviews
- Physical Sciences › Computer Science › Information Systems › Data Quality and Management
- Social Sciences › Library and Information Sciences › scientometrics and bibliometrics research
- Keywords
- EndNote, bibliographic database searching, citation screening, sensitivity and specificity, reference management software
- MeSH terms
- Algorithms, Humans, Publications, Records, Sensitivity and Specificity, Reproducibility of Results, Databases, Bibliographic, Information Storage and Retrieval
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Better duplicate detection for systematic reviewers: evaluation of Systematic Review Assistant-Deduplication Module. https://app.cochat.ai/papers/10.1186/2046-4053-4-6
- The paper
- Rathbone, J., Carter, M., Hoffmann, T., & Glasziou, P. (2015). Better duplicate detection for systematic reviewers: evaluation of Systematic Review Assistant-Deduplication Module. Systematic Reviews. https://doi.org/10.1186/2046-4053-4-6
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.