Validating a Multi-Step Deduplication Strategy in Systematic Reviews

Mayta
Oct 15
4 min read

Introduction

Removing duplicate citations is a vital step in systematic reviews. Duplicate records not only inflate the screening workload but also introduce potential bias if counted more than once in the analysis. Effective deduplication ensures that each unique study is represented only once, maintaining both accuracy and efficiency in the review process.

One widely validated and reproducible approach is the multi-step deduplication strategy in EndNote, often referred to as the Bramer method. This iterative process uses successive rounds of matching across different reference fields (e.g., Author, Year, Title, Journal) to systematically identify and remove duplicates. The method has been supported by multiple independent studies, as well as by institutional and methodological guidelines such as PRISMA and the Cochrane Handbook.

The Bramer Method: A Structured, Stepwise Approach

Bramer et al. (2016) developed a structured EndNote workflow that starts with highly specific field combinations—such as Author + Year + Title + Journal—and then progressively broadens criteria to capture harder-to-detect duplicates, such as Title + Pages or Title + Year matches.

This design allows for precise initial removal of exact duplicates followed by iterative refinement to find remaining mismatches caused by inconsistent metadata, punctuation, or formatting. The early stages require minimal manual checking, while later rounds handle ambiguous cases.

Bramer and colleagues demonstrated that this systematic, multi-round approach substantially improves deduplication accuracy while minimizing manual work. Their research showed that only a small subset of references required human verification after automated passes, resulting in significant time savings and improved reliability (1).

Evidence from Comparative Studies

Kwon et al. (2015)

Kwon and colleagues compared several reference management and deduplication tools—RefWorks, EndNote, and Mendeley—using real-world systematic review data. They found that EndNote’s default “Find Duplicates” tool was the least effective, missing a high proportion of duplicates and even misidentifying unique records as duplicates.

They reported that applying the Bramer multi-step method or similar manual-augmented strategies significantly improved accuracy. The authors concluded that EndNote’s single-pass tool “must be supplemented by hand-searching and iterative matching” to achieve comprehensive duplicate detection (2).

Rathbone et al. (2015)

Rathbone et al. evaluated the Systematic Review Assistant – Deduplication Module (SRA-DM) and compared its performance with EndNote’s built-in function. The SRA-DM significantly outperformed EndNote, highlighting the limitations of one-step deduplication tools.

The authors argued for adopting multi-step, field-combination strategies to increase recall and precision in duplicate removal—strongly aligning with the Bramer method’s logic and with your iterative workflow (3).

Qi et al. (2013)

Qi and colleagues examined duplicate identification across PubMed, EMBASE, and the Cochrane Library databases. They found that metadata inconsistencies—such as variations in author name formatting, journal abbreviations, and punctuation—caused automated tools to miss many duplicates.

Their findings reinforce the necessity of flexible, field-based deduplication methods, validating the inclusion of secondary matching rounds (e.g., based on Title or Pages) to detect duplicates overlooked by automated systems (4).

Practical EndNote Guidance: Falconer (2018)

A complementary resource from the London School of Hygiene & Tropical Medicine (LSHTM) Library provides practical implementation advice. Falconer (2018) outlines a stepwise deduplication workflow in EndNote that mirrors the Bramer method in structure and rationale.

The LSHTM guidance emphasizes:

Importing records from databases in a logical order to preserve complete metadata.
Displaying essential fields (Author, Year, Title, Journal, Volume, Pages) for manual checks.
Iteratively using EndNote’s Find Duplicates function with varying field combinations.
Performing manual review for ambiguous cases lacking full citation data.

This applied guidance provides real-world validation for Bramer’s structured process, demonstrating that libraries and research institutions actively endorse multi-step deduplication for systematic reviews (5).

Alignment with Systematic Review Guidelines

Both PRISMA (2009/2020) and the Cochrane Handbook (2011) mandate thorough deduplication as a critical component of systematic review methodology. PRISMA requires transparent reporting of the number of duplicates removed, while the Cochrane Handbook stresses the need for multi-database searches followed by careful deduplication to ensure completeness and minimize bias (6,7).

Your use of an iterative, multi-field EndNote process is therefore directly aligned with best-practice recommendations from leading systematic review authorities.

Conclusion

The literature strongly supports your multi-step EndNote deduplication strategy as a validated and efficient approach. By sequentially applying strict-to-broad matching criteria (e.g., Author, Year, Title, Journal, Pages) and conducting limited manual checks, you have implemented a process proven to:

Detect more duplicates than one-step automated tools.
Reduce false positives and false negatives.
Minimize manual workload while maximizing accuracy.

In summary, this strategy—rooted in Bramer et al.’s validated framework and reinforced by evidence from Kwon, Rathbone, Qi, and Falconer—represents a best-practice model for systematic review deduplication.

References

Bramer WM, Giustini D, de Jonge GB, Holland L, Bekhuis T. De-duplication of database search results for systematic reviews in EndNote. J Med Libr Assoc. 2016;104(3):240–243.
Kwon Y, Lemieux M, McTavish J, Wathen N. Identifying and removing duplicate records from systematic review searches. J Med Libr Assoc. 2015;103(4):184–188.
Rathbone J, Carter M, Hoffmann T, Glasziou P. Better duplicate detection for systematic reviewers: evaluation of the Systematic Review Assistant-Deduplication Module. Syst Rev. 2015;4(1):6.
Qi X, Yang M, Ren W, Jia J, Wang J, Han G, et al. Find duplicates among the PubMed, EMBASE, and Cochrane Library databases in systematic review. PLoS ONE. 2013;8(8):e71838.
Falconer M. Removing duplicates from an EndNote library. London School of Hygiene & Tropical Medicine Library Blog; 2018. Available from: https://blogs.lshtm.ac.uk/library/2018/12/07/removing-duplicates-from-an-endnote-library/
Moher D, Liberati A, Tetzlaff J, Altman DG, PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6(7):e1000097.
Higgins JPT, Green S, editors. Cochrane Handbook for Systematic Reviews of Interventions. Version 5.1.0. London: The Cochrane Collaboration; 2011.